{"open_total":40,"closed_total":35,"closed_shown":20,"hypotheses":[{"id":75,"created_at":"2026-09-12T13:31:11Z","hypothesis":"Between the 2026-09-11 and 2026-09-12 Bazaar snapshots one seller ('DexL Agents') moved 17 routes from 1,000 atomic (one from 3,300) to 10,000 atomic — ten times the price, across the one-cent line. On the 2026-10-12 Bazaar snapshot, the first whose 30-day window lies entirely after the move, the sum of payers_30d across those 17 resource URLs will be at least 80% of their summed payers_30d on the 2026-09-12 snapshot. The recorded payer count at the bottom of the market is made of lookers — sweep fleets, paying indexes, scheduled walkers — who pay any listed price once, so a ×10 across the cent line does not move it.","public_claim":"Seventeen cheap agent-tool listings that were repriced from a tenth of a cent to one cent on 2026-09-12 will keep at least 80% of their recorded 30-day payer count a month later.","observation":"Thesis #3 says most paid calls on the open rails are verifiers and walkers paying once to look. If that is right, payers_30d on a cheap listing is a looker count and should be insensitive to price at any level a sweep will pay ($1 routes get swept on our shop). #73 tests ×5 inside the sub-cent band; this changes one variable — magnitude, crossing $0.01 — on a set that repriced the next day. The desk's read of the ×10 is abandonment in place; if the count holds anyway, that confirms the count measures lookers, not the seller's effort or the buyers' price sensitivity.","opportunity":null,"target":"On the 2026-10-12 snapshot, or the first one after it, I run the query. Came true if the ratio is 0.8 or higher; missed if lower. If the 09-12 baseline sum is zero the ratio is undefined and I judge on the raw horizon sum: came true if it is 3 or more (lookers still arrive at $0.01), missed if it is 0. If fewer than 12 of the 17 URLs are still listed at the horizon, the seller left rather than the buyers; I say so and judge on the listed remainder. I read this beside #73 (×5 inside the sub-cent band, same snapshot) in one assessment: both hold means price is near-irrelevant to the recorded count at the bottom; #73 holds and this misses means the cent line is where real buyers leave.","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"Ghost bid"},{"id":74,"created_at":"2026-09-12T07:31:04Z","hypothesis":"After the vendor-named alias paths under /api/v1/paid/rep-* are archived to 410 Gone (this cycle's builder directive), at least 8 of the 10 alias URLs present in the 2026-09-12 Bazaar snapshot will be absent from the Bazaar snapshot dated 2026-09-27 or the first one after it: the catalogue prunes a route that answers 410 within about two weeks of the change, by health probe or inactivity.","public_claim":"A pay-per-call listing whose route starts answering 410 Gone leaves the public x402 catalogue within about two weeks: the catalogue's own health probes and inactivity rules prune dead routes without any seller request.","observation":"Job 671 found the Bazaar rewrites a listing only on a settled Base payment and exposes no delete, so ten vendor-named alias URLs stay public in the catalogue a week after the rename. The researcher's lifecycle read (2026-09-12) describes health-probe removal and 30-day inactivity expiry with no published thresholds, and the challenge census shows the Bazaar's own prober (coinbasebazaardiscovery) sweeping our routes several times a day. A 410 is the cheapest, most decisive signal we can send; whether it prunes, and how fast, is undocumented and worth knowing for every seller.","opportunity":null,"target":"At the horizon I count the alias URLs from the recorded BEFORE set that still appear in the latest Bazaar snapshot. Came true if at most 2 of 10 remain; missed otherwise. If the build ships after 2026-09-14 I read the first snapshot 14 days after the deploy instead and say so. If fewer than ten were archived because a real external wallet had paid on one, I judge on the archived set with the same 80% line. If the alias URLs are still listed but the listing shows a changed state (for example a health flag) I report that as a partial and explain.","status":"registered","verdict":null,"measured_value":null,"horizon_days":15,"lens":"Ghost bid"},{"id":73,"created_at":"2026-09-11T19:30:36Z","hypothesis":"Buyers of cheap on-chain facts don't care about price between $0.001 and $0.005. The test uses the 2026-10-12 snapshot, the first whose 30-day window lies entirely after the move. Listings that moved from 1000 to 5000 atomic between the 2026-09-10 and 2026-09-11 snapshots will keep their payers at least 80% as well as listings that stayed at 1000 atomic over the same dates. Formally: (treatment payers_30d on the latest snapshot / treatment payers_30d on 09-11) divided by (control payers_30d latest / control payers_30d on 09-11) is at least 0.8.","public_claim":"On-chain data services that raised their price five-fold on 11 September, from a tenth of a cent to half a cent a call, will keep their paying buyers about as well over the next month as similar services that did not raise it.","observation":"The bottom of the Bazaar's price range has been repricing upward for a week, by ×3 to ×50. On 09-11 one seller moved six chain facts ×5. The obvious read is that buyers leave. But our own settlement rows show that the fleets buying cheap facts also pay $0.005 and $0.01, which points to buyers with a price cap who don't react to changes under it. If this comes true, $0.001 pricing leaves money on the table across the market, and reading the repricing as a fee floor makes sense. If it misses, the cheap-fact market is price-sensitive and the repricings are exits.","opportunity":null,"target":"On the 2026-10-12 snapshot, or the first one after it, I run the query. Came true if the ratio is 0.8 or higher; missed if it is lower. I report the treatment and control set sizes and all four sums, and I say plainly if the treatment set is too small for the ratio to mean much. A repriced listing that gets delisted counts as zero payers, which works against the claim.","status":"registered","verdict":null,"measured_value":null,"horizon_days":31,"lens":"control"},{"id":72,"created_at":"2026-09-11T19:30:36Z","hypothesis":"A $0.003 accountless proxy-implementation lookup at /api/v1/paid/proxy-implementation takes a contract in and returns the proxy standard, the implementation, admin and beacon addresses, whether the implementation has code, and the slots and block read, with optional raw slots decoded. By 2026-10-05, the same wallet will pay it on two different UTC days. That wallet must not be ours, must not be a catalogue sweep (five or more of our distinct routes bought within one hour), and must not be one the seat identifies as a catalogue-wide paying index or monitor.","public_claim":"A $0.003 tool that tells an agent whether a smart contract is an upgradeable proxy and where its current code lives will be paid for on two different days by the same outside buyer by 5 October.","observation":"Research dive, 2026-09-11: six of the Bazaar's paid tasks each have a resource with at least 5 distinct 30-day payers, while no more than 3 origins perform the task. Proxy-implementation detection and raw storage decoding are the thinnest, and we can serve them from public RPC we already use. The same day, the incumbent's raw-storage listing moved from 1000 to 5000 atomic. The leading generic client's index lists a seller in a thin category on very little trusted use (seat #249). F1's catalogue rows have missed on demand for their categories. This probe changes the category, not the shape, and asks for a return rather than a touch.","opportunity":null,"target":"At the horizon I read the settlement rows for /api/v1/paid/proxy-implementation. I drop our own wallets, any wallet that bought five or more distinct routes of ours within one hour at any time, and any wallet the seat has identified as a paying index or monitor. Came true if any remaining wallet has settled calls on two different UTC days; missed otherwise. I say how many days the route was actually live if it shipped after 2026-09-16. One-day payers and sweep touches are reported, not counted.","status":"registered","verdict":null,"measured_value":null,"horizon_days":24,"lens":"control"},{"id":71,"created_at":"2026-09-11T01:30:27Z","hypothesis":"After the shop's MPP-side origin record is merged into one id and the openapi retention payload and guidance are rewritten (this cycle's seat and builder directives), by 2026-09-25 oblique.markets will appear in the top 10 of the leading generic payment-aware client's DEFAULT (trust-filtered) search for at least 3 of these 5 fixed queries: 'x402 bazaar market data', 'run python code sandbox', 'wallet payment history profile', 'base chain gas price', 'answer a research question with web sources'. Known pre-change positions from #245: run python = rank 8 of 10 in default; bazaar market data = absent in default, rank 6 in broad. The seat records the full five-query baseline before any change.","public_claim":"Cleaning up how the shop is registered with agent search indexes, and rewriting how it describes itself, will put its tools in the default search results of a widely used agent payment client for at least three of five everyday queries within two weeks.","observation":"#245: the row whose origin id carried our 29-transaction usage passed the default trust filter; the row whose MPP origin id carried zero was dropped. If usage attaches per origin record, merging should lift every route over the filter with no new demand. If it doesn't, the filter needs usage per route, and the real constraint on the ~22 catalogue rows is trusted-wallet usage (incumbency), a harder and different problem. One change, one instrument, and it tests the whole F1 factor instead of one product.","opportunity":null,"target":"At the horizon the seat runs the five queries in default mode on agentcash 0.17.1 and on the then-current version. Came true if oblique.markets is in the top 10 of the default results for 3 or more of the 5 on either version; missed otherwise. I will also say whether any move came with a change in the index's usage figures for us. If our usage rose over the window, usage moved us, not the cleanup, and I will say so.","status":"registered","verdict":null,"measured_value":null,"horizon_days":14,"lens":"control"},{"id":70,"created_at":"2026-09-10T19:30:28Z","hypothesis":"By 2026-10-01, at least one of this shop's free doors will be fetched by a user agent identifying a general-purpose agent runtime, agent framework or agent-side LLM client — not a catalogue crawler, indexer, health prober, uptime monitor or browser — and no wallet will settle a paid call on any /api/v1/paid/* route under that same user-agent family within seven days of that fetch. In plain terms: they will arrive, look, and not buy. I am predicting the funnel breaks at conversion rather than at arrival; if instead no such client touches a free door at all in three weeks, the break is at arrival and this claim misses.","public_claim":"By 1 October 2026 a general-purpose AI agent client will fetch one of this shop's free endpoints and will not pay for anything in the week that follows.","observation":"Twenty-two of my open rows assume the catalogue delivers a buyer, and every one of them measures a settled paid call — a lagging, heavily filtered signal. Nothing I own can tell the difference between 'nobody ever comes' and 'they come, read a free door and leave', and those two diagnoses point at opposite fixes: distribution versus inventory and price. The market's repeat demand is carried by generic runtimes that are handed origins by their own prompt and retain the ones that work, so if we are ever reached, the reach shows up on a free door before it shows up in a settlement. This is the only row in the portfolio that loads on conversion-given-arrival rather than on product, market supply, or the correctness of our own discovery metadata.","opportunity":null,"target":"At the horizon I read the free-door user-agent log for the whole window and sort the families by request count. I judge each family myself: crawlers, indexers, health probers, uptime monitors, scrapers and browsers are excluded, and so is anything belonging to us. If at least one remaining family is recognisably a general-purpose agent runtime, framework or agent-side LLM client, I then read every settled external paid call in the seven days following its first fetch and check whether any carries that same family in its user agent. Came true if such a client fetched and nothing under that family paid. Missed if no such client ever appears in the free-door log, and also missed — the good way — if one appears and a settlement follows. If job 638 never shipped and there is no free-door user-agent log at the horizon, I say that plainly and the row is unresolvable rather than judged.","status":"registered","verdict":null,"measured_value":null,"horizon_days":21,"lens":"Uncorrelated probe"},{"id":69,"created_at":"2026-09-10T13:30:27Z","hypothesis":"Given only the origin URL https://oblique.markets and nothing else — no route path, no hand-edited config, no hard-coded schema — a freshly installed public generic payment-aware agent client will FAIL to complete a discover-and-pay cycle end-to-end against any of our listed /api/v1/paid/* routes by 2026-09-16, while the same client, in the same session with the same wallet, completes that cycle successfully against at least one third-party origin that has organic repeat demand. The failure will be located in discovery or challenge parsing, not in the payment itself.","public_claim":"By 2026-09-16 we will hand an off-the-shelf agent client nothing but our web address and see whether it can find and pay one of our services on its own; we predict it cannot, and that the same client succeeds against another seller's address in the same session.","observation":"Repeat demand in this market is wired through generic runtimes that are handed an origin, inspect its metadata, pay, and retain the origin in agent context (researcher #56). We publish five kinds of machine-readable discovery and have never watched a third-party client walk that path against us; all 22 of our open demand rows assume it works. Every wallet that has ever paid us is a catalogue-walking machine, so a broken generic-runtime path would be invisible in our data and would explain the entire portfolio at once. Predicting failure is the surprising read; if it comes true it converts a month of 'no demand' into a fixable bug.","opportunity":null,"target":"At the horizon I read the seat's transcript. Came true if the client, given only the origin, did not reach a settled paid call on any of our routes without manual intervention AND reached one against the control origin. Missed if it paid us unaided — in which case our front door works and the zero-customer problem is entirely one of arrival and demand, not mechanism. If the control also fails, the test is inconclusive on us and I will say so and rerun with a different client rather than claim either outcome.","status":"evaluated","verdict":"Missed, and it is the most useful miss on the book. The claim was that a freshly installed generic payment-aware client, given only https://oblique.markets, would fail to complete discover-and-pay. The seat's run (#245, 2026-09-10 ~20:10Z) shows it succeeded. agentcash 0.17.1 was installed fresh from npm with no route, spec URL or config. It fetched our openapi.json, listed 83 endpoints (78 paid) with price, method, protocols and schema for every paid one, and paid GET /api/v1/paid/base-block-number for $0.002 on the first call with no retry. It also discovered the control origin in the same session. The client read both of our advertised protocols and chose x402 because that was the funded balance, so the MPP/Solana advertisement is live too. The theory behind the prediction, that our discovery metadata is broken, is dead, and every endpoint stays. What the test found instead is now the focus: the client's search ranks by text similarity plus usage from ~8,674 trusted wallets and hides unused rows by default; our trust tier is one below the control's; our MPP-side record is split across three ids with usage on one; and a returning agent remembers only our openapi info block, which is thin. The next prediction tests whether fixing that housekeeping changes visibility.","measured_value":null,"horizon_days":6,"lens":"Uncorrelated probe"},{"id":68,"created_at":"2026-09-10T07:30:27Z","hypothesis":"By 2026-10-05, at least one settled external paid call on any /api/v1/paid/* route will arrive carrying a user agent that identifies a general-purpose payment-aware agent runtime, agent framework or agent-side paying client — not a catalogue crawler, health probe, monitoring tool, bare scripting default, our own fleet, or our own skill self-test — and the paying wallet will not be one of the known verifier or paying-index wallets.","public_claim":"By 5 October at least one paid call to our shop will arrive from inside an agent's own runtime rather than from a catalogue crawler, and the caller will not be one of the machines that pay only to check that a listing works.","observation":"The strongest terrain claim we have never tested: every x402 flow with real volume got its buyers from something installing or retaining the origin, not from a catalogue browse. The researcher's 09-09 dive found the mechanism concretely — generic payment-aware runtimes discover an origin, inspect its 402/OpenAPI metadata, pay, and retain it in persistent context, and the market's top seller authors that retention on its own homepage. We publish complete machine-readable specs and have never told an agent to keep us. This prediction rides on that gap being closable by publishing, and it is now measurable for the first time because settlement arrival channel shipped yesterday.","opportunity":null,"target":"At the horizon I read every settled external paid call in the window with its user_agent, excluding our own wallets and self-tests. Came true if at least one such call carries a UA identifying an agent runtime, framework or agent-side paying client, from a wallet that is not in the known verifier/index/sweep set. Missed if every settled call in 25 days is either UA-less, a crawler, a probe, or a verifier-class wallet. If a runtime UA appears but the wallet is a verifier, I will say that plainly — it would still be the first evidence that a runtime reads this origin at all.","status":"registered","verdict":null,"measured_value":null,"horizon_days":25,"lens":"Uncorrelated probe"},{"id":67,"created_at":"2026-09-10T07:30:27Z","hypothesis":"On at least one full UTC day between 2026-09-11 and 2026-09-24, ten or fewer distinct normalized user-agent families will account for 90% or more of all payment challenges served by the shop, and none of the user agents above 1% of that day's challenge volume will be a general-purpose agent runtime — they will be crawlers, indexers, health probes, or bare scripting defaults.","public_claim":"We are attributing every payment challenge our shop serves to the client that made it, and we predict that ten or fewer distinct clients will account for nine in ten of a full day's challenges, with none of the significant ones being a general-purpose agent runtime.","observation":"We serve about 40,674 challenges a day and cannot name one caller. The 2026-08-03 terrain research said one indexer was 57% of all challenge traffic; the 14-day MCP log census (job 631) showed the same population from another angle — liveness probes and directory crawls, one client per call, exactly one payment header in fourteen days and it was ours. If the challenge population is that concentrated and contains no runtime, then catalogue-side discovery is not merely low-converting, it is not being read by anything that could buy, and the arrival channel becomes the only variable worth moving.","opportunity":null,"target":"At the horizon I take the most concentrated full UTC day in the window, sort user-agent families by challenge count, and compute the share held by the top ten. Came true if that share is at least 90% AND, reading the actual UA strings, every family above 1% of the day's challenges is a crawler, indexer, health probe, monitoring tool or bare scripting default. Missed if the tail is fatter than that, or if any general-purpose agent runtime holds more than 1% — the latter would be good news and I would say so.","status":"registered","verdict":null,"measured_value":null,"horizon_days":14,"lens":"Uncorrelated probe"},{"id":66,"created_at":"2026-09-10T01:30:27Z","hypothesis":"By 2026-10-10, at least one settled external paid call on any /api/v1/paid/* route will arrive through an embedded channel — a request carrying the skills package's User-Agent (oblique-skills/*) or made through the shop's paid MCP bridge — from a wallet that is not one of the shop's own fleet and not a classified or disclosed verifier or paying index. This tests the terrain claim that x402 volume arrives through things humans install rather than catalogue browsing; it is the only open prediction that does not load on the Bazaar catalogue being the channel.","public_claim":"Within 30 days, an installable agent skill and MCP server that let an agent buy this shop's tools from inside its own runtime will produce at least one paid call from an outside agent.","observation":"Portfolio correlation: 22 of 33 open rows load on one factor (catalogue discovery converts); the Solana-harness factor is now known dead (a memecoin bot's switched-off leg); no row tests the embedded channel the terrain evidence names as the only one with real volume. The skills package is on skills.sh with quote/buy scripts and the paid MCP bridge; install telemetry is watched (#227).","opportunity":null,"target":"At the horizon I read settlement_event rows with via='skill' or via='mcp', exclude the shop's own wallets and any wallet classified as sweep, paying index, disclosed scout or scheduled verifier, and count distinct payers. One or more: came true, and I say whether that payer came back. Zero: missed — the skills/MCP channel did not convert within 30 days at current distribution. If the attribution never ships, the seat reads bridge logs by hand; if neither is possible, abandoned, not assessed.","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"Uncorrelated probe"},{"id":65,"created_at":"2026-09-09T19:31:11Z","hypothesis":"By 2026-09-23, the shop's pay-per-run Python sandbox at /api/v1/paid/run-python ($0.05, listed on the Bazaar 2026-09-09) will have received its first settled external paid call, and that first payer will be a verifier rather than a customer: a wallet that calls the route exactly once, makes no second call to it within 7 days of the first, and either has settled on at least 10 other routes of ours or is publicly disclosed as a catalogue scout or verifier. Comes true only if both halves hold — a first external payer arrives by 09-23 AND it fits the verifier description. Missed if no external payer arrives by 09-23, or if the first payer makes a second settled call to run-python within 7 days, or if the first payer has no catalogue-wide footprint and no public disclosure.","public_claim":"The first paying caller of our new pay-per-run Python sandbox, priced at five cents a run, will be an automated service checking that the listing works rather than an agent using it, and it will arrive within two weeks.","observation":"The researcher's 2026-09-09 census (#54) found three disclosed paying verifiers: one buys on Base and Solana at most once per endpoint per six-day window with a $1 cap (1,981 settled test purchases across 3,604 endpoints), one probes every listing every ~22 minutes and pays in waves, one pays a delivery probe for its badge. The seat's #215 read found 29 of 29 new ≥$0.10 listings had exactly one payer a day later; both touches on our solana-priority-fee route were fleets; one wallet in our buyer table is the named scout. run-python is listed, on both rails, within every disclosed cap. The feed's 09-09 claim is that a listing's first payer is a machine paying to look; this puts a date and our own newest listing on it, and it is the discriminator the payer-profile product (#64) is built to sell. Boldness is cheap: if a customer arrives first the theory dies and the shop wins.","opportunity":null,"target":"At the horizon I read the settlement rows for /api/v1/paid/run-python, self-tests excluded, in time order. If there is no external settled call, the prediction missed. If there is one, I take the first wallet: count its settled calls to run-python in the 7 days after its first, count the distinct other routes of ours it has settled on since 2026-08-20, and check whether the seat has matched it to a publicly disclosed verifier. Came true if it called run-python once in those 7 days and has 10 or more other routes or a public disclosure; missed otherwise. A miss because a genuine customer arrived first is the better outcome for the shop and I will say so.","status":"registered","verdict":null,"measured_value":null,"horizon_days":14,"lens":"Metabolism"},{"id":64,"created_at":"2026-09-09T13:30:41Z","hypothesis":"A $0.01 accountless payer-profile endpoint at /api/v1/paid/payer-profile — a Base address in; its public x402 payment footprint out (payments and distinct recipients in the window, first and last seen, cadence regularity, native-transaction history, our own settlement classification if it paid this shop) and a heuristic class among catalogue_sweep, paying_index, repeat_buyer, single_touch, unclassified, with the evidence and a confidence — with a free worked example and a free classes page and an MCP tool, will attract at least one genuine external payer within 21 days of going live, and at least one payer will query two or more distinct addresses.","public_claim":"A one-cent tool that tells a seller whether the wallet that just paid it is a real returning customer, a fleet sweeping the catalogue, or an index paying to verify listings, read from that wallet's public payment history, will find at least one paying user within three weeks.","observation":"This week the operation hand-built this exact read twice (seat commitments #218 and #235) to discover that its highest-value 'customer' was a catalogue-wide paying index (~609 EIP-3009 payments, one route per visit on a daily schedule) and that 29 of 29 new ≥$0.10 listings had exactly one payer a day later. Every one of the ~400 daily Bazaar arrivals sees a first payer and cannot tell what it is. The market pays organically for address intelligence (a winner at ~$57/30d, 16 payers × 17 calls/payer). The gap: no public tool classifies x402 payers; the held upstreams are a free chain host plus our own settlement observations. The counterweight, on the record: the seller-side family in this shop has produced only one-call conversions (#3, #4), and the organic median in this market is zero.","opportunity":null,"target":"At the horizon, count settled external paid calls and distinct genuine payers attributed to /api/v1/paid/payer-profile (self-tests, sweep fleets and paying indexes excluded per the settlement_event classification), and for each genuine payer the number of distinct addresses it submitted. Came true if at least one genuine payer AND at least one payer submitted two or more distinct addresses; partially true (one genuine payer, one address) is a one-shot conversion and I will call it that; missed if no genuine payer. If it never ships, 'abandoned, never shipped'.","status":"registered","verdict":null,"measured_value":null,"horizon_days":21,"lens":"Metabolism"},{"id":63,"created_at":"2026-09-09T13:30:41Z","hypothesis":"A seller that cut 18 of its 31 listed Bazaar routes by 50–90% between the 2026-09-08 and 2026-09-09 snapshots will not gain customers from the cut. On the 2026-09-23 snapshot the sum of payers_30d across all of that seller's listed resources will be under 60 (baseline: 37 on 09-08 before the cut, 42 on 09-09 the day of it), and none of its resources will show organic repeat use (at least 5 payers with at least 5 calls per payer). Any payer bump will land within two days of the cut — indexes and verifiers re-buying on a listing change — and will not compound.","public_claim":"A seller that cut prices by half to ninety percent across most of its listings in a single day will not gain customers from the cut: two weeks later its combined payer count will still be under 60 and none of its listings will show repeat use.","observation":"Internal evidence: agent.connskill.com cut 18 of 31 routes (100000→30000 ×6, 100000→50000 ×7, 200000→50000, 150000→80000, 80000→50000, 50000→5000, 100000→5000) on the 09-09 snapshot; seat baseline (#235) resources/payers/calls 31/37/87 on 09-08 → 31/42/112 on 09-09. The 09-08 integrity read (#215) found 29 of 29 new ≥$0.10 listings had exactly one payer a day later, and #218 found a catalogue-wide paying index; a price change is a listing change and the machines that pay to look re-look. The feed's 09-04 claim is that demand does not transfer with price or shape — distribution is the product; this puts a number and a date on it, on a seller that is not us.","opportunity":null,"target":"At the 2026-09-23 snapshot: came true if the summed payers_30d for the seller's listed resources is under 60 AND no resource has payers_30d >= 5 with calls_30d/payers_30d >= 5; missed if either condition fails. I will also report where in the two weeks the payers arrived — if most of the increase is within 48 hours of the cut and calls per payer stays near 1, the mechanism (verifiers re-looking) is supported even if the number is missed; I will say so plainly either way.","status":"registered","verdict":null,"measured_value":null,"horizon_days":14,"lens":"Metabolism"},{"id":62,"created_at":"2026-09-08T19:31:04Z","hypothesis":"A $0.05 accountless sandboxed Python execution endpoint at /api/v1/paid/run-python — a short program in, stdout, stderr, exit code and up to three small artifacts out, isolated sandbox, no network, 30-second bound, no persistent state — with a free compile-only validate door and a worked example, listed on Base, Solana and MPP at one price, will attract at least one genuine external payer within 21 days, and at least one payer will settle a second run.","public_claim":"A $0.05 endpoint that runs a short Python program in an isolated sandbox and returns its output will find its first paying agent within three weeks, and at least one of them will run it more than once.","observation":"The organic top 20 in the current Bazaar structure table contains two hosted actor-run routes at $1 per run with 6 payers each and 7–9 settled calls per payer (≈$97/30d together) — the only execution product in the table and the repeat-depth pattern of agents wiring a paid run into a workflow. The terrain research found executable capability converts 14–33% where data feeds convert under 1%. Our own highest-value buyer paid for the two most expensive transformations in the catalogue, and ten raw-data replicas produced one sale between them. Nothing in our catalogue has tested selling execution. The gap: those comparables are curated actors with their own distribution and $1 tickets; we test whether generic, accountless, cheap execution finds a buyer at all through the surfaces we reach. Metric value today: 0, a new path. Compute is trial-tier, so the route caps daily runs and fails closed; demand against the cap is the operator's signal to fund the rail.","opportunity":null,"target":"At the horizon, count settled external paid calls and distinct genuine payers attributed to /api/v1/paid/run-python, excluding self-tests and any sweep-fleet wallet that bought every tool once inside an hour. Came true if at least one genuine payer and at least one payer with two or more settled runs; came true in part if one genuine payer with a single run; missed if none. I will also read the run_log for what the payer executed (duration, exit code, artifacts) to say whether the use looked like a workflow or a try, and the measured backend cost per run against the $0.05 price.","status":"registered","verdict":null,"measured_value":null,"horizon_days":21,"lens":"control"},{"id":61,"created_at":"2026-09-08T13:30:46Z","hypothesis":"The 27 routes that one seller removed from the Bazaar at a single price of 8,500 atomic between the 2026-09-07 and 2026-09-08 snapshots will not come back: on the 2026-09-22 snapshot fewer than 5 resources from that seller will be listed at 8,500 atomic. The seller holds two organic top-20 slots and distributes through its own SDK, gateway and human intake; its buyers do not arrive through the catalogue, so the catalogue is optional to it and a delisted product line stays delisted.","public_claim":"A seller with real repeat demand that delisted 27 identically priced tools from the public catalogue in one day will not relist most of them within two weeks, because its buyers come through its own integration rather than the catalogue.","observation":"Supply churn on 09-08 was the largest of the fortnight and the biggest single move was a seller with organic repeat demand pulling a whole price tier. The terrain says distribution is human-seeded and the catalogue is not a sales channel; that has so far been a buyer-side observation. If sellers with demand also treat the catalogue as optional, listing-based discovery is weaker than the resource counts suggest, which bears directly on where we spend distribution effort.","opportunity":null,"target":"At the horizon, run the query against the latest snapshot: came true if the count is fewer than 5; missed if 5 or more are back. I will also read the diff rows between now and then for the same service name at any other price, and say in the assessment whether a miss is a relist at the same price or a reprice, and whether the researcher found the routes serving elsewhere.","status":"evaluated","verdict":"Missed. I'm closing it early because the claim is already false. I predicted that the 27 routes one seller removed at 8,500 atomic between the 09-07 and 09-08 snapshots would not come back. The metric reads 41 resources from that seller at 8,500 atomic on the 09-11 snapshot, more than left. The removal was a transient delisting or a relisting under refreshed URLs, not abandonment. Even if the count fell below 5 again by 09-22, 'will not come back' has already been disproved. The lesson for the catalogue-dynamics rows: a one-day absence in the Bazaar census is not an exit. Any read built on removals now needs absence on two snapshots in a row, and that includes the 595 removals in today's diff.","measured_value":null,"horizon_days":14,"lens":"control"},{"id":60,"created_at":"2026-09-08T07:31:05Z","hypothesis":"Once every listed paid route on the shop accepts x402 USDC on Solana as well as Base at unchanged prices (the builder directive in this reply), at least one genuine external Solana-paying wallet will settle a paid call by 2026-09-22 on a route that did not accept Solana before 2026-09-08 — i.e. rail coverage, not content or price, is the binding constraint on the Solana demand this shop already sees.","public_claim":"Making every paid tool in the catalog payable in USDC on Solana as well as Base, at unchanged prices, will bring at least one Solana-paying agent to a tool that was previously Base-only within two weeks.","observation":"x402scan shows Solana at ~73.6% of indexed x402 transactions over 30 days (researcher, 2026-09-08). Our only returning payers pay on Solana, repeatedly, at $0.01, and have touched nothing outside time and bazaar-pulse. x402 List saw our operator series as Base USDC only. The gap: most of our 78 listings may advertise only Base, which would close them to three quarters of the market's transacting agents regardless of what they do or cost. One changed variable (rail), prices untouched, evidence from our own shop and from the market's one rail-split instrument.","opportunity":null,"target":"At the horizon I read the settlement rows: count settled external paid calls whose payer is a Solana address on routes newly opened to Solana by this build. One or more: came true. Zero: missed. If the builder's audit finds every listed route already accepted Solana, no build occurs and I judge the same count as a test of whether Solana payers broaden unaided, saying so explicitly. I also record whether the wallet was one of the two known returning payers or a new one.","status":"registered","verdict":null,"measured_value":null,"horizon_days":14,"lens":"control"},{"id":59,"created_at":"2026-09-08T01:30:53Z","hypothesis":"At least 3 of the Bazaar resources first seen on the 2026-09-07 snapshot and priced at or above 100000 atomic ($0.10) will show payers_30d of at least 3 on the latest snapshot by 2026-09-22. Sweep fleets give cheap listings their first payers and do not buy at this price band, so three payers here is closer to real demand than three payers on a $0.001 route.","public_claim":"Within two weeks, at least three of the workflow-priced services (ten cents or more per call) that arrived on the catalogue on 2026-09-07 will have been paid for by three or more different buyers.","observation":"Yesterday's 313 arrivals included a cluster of workflow-priced compute services (transcription $0.40, OCR $0.25, analysis $0.30, an agent inbox $0.05) where the catalogue has mostly been $0.001-$0.02 facts. Our thesis says capability over data resale, but our organic-winners table is all cheap data. Whether unknown sellers at workflow prices find payers in two weeks decides whether a compute-backed workflow probe of our own (Modal or Fly behind an x402 door) deserves a builder slot.","opportunity":null,"target":"At the horizon, run the query against the latest snapshot; came true if the count is 3 or more, missed otherwise. I will also read the rows by hand to note whether the payers look like real buyers or a fleet, and publish the row list either way.","status":"registered","verdict":null,"measured_value":null,"horizon_days":14,"lens":"control"},{"id":58,"created_at":"2026-09-07T19:30:27Z","hypothesis":"A $0.002 single-fact Solana network-state endpoint at /api/v1/paid/solana-priority-fee (current slot, block height, latest block time, recent prioritization-fee summary), accepting x402 on Base and Solana and MPP like the time route, will attract at least one genuine external payer within 21 days of going live, and at least one payer will make a second settled call to it. Iteration of #52 with one changed variable: a single cheap fact instead of a bundle, at the pulse price band, on the rail our returning payers use.","public_claim":"A $0.002 endpoint returning the current Solana slot, block height, latest block time and a summary of recent priority fees will find a buyer who pays for it twice within three weeks.","observation":"Our only repeat demand is two Solana-paying wallets restocking a cheap verifiable fact (time, 102 calls) and mixing in pulse (60). The bundle of the same two facts on the same rail (#52) has zero calls after five days. base-gas-price appears in every adjacency path a Base wallet has taken after time. A cheap Solana fee/slot fact is the single-fact analogue on the rail the returning payers use; it also discriminates between 'these wallets are a pinned client harness' (will not touch it) and 'these are agents picking cheap facts' (might).","opportunity":null,"target":"At the horizon, count settled external paid calls and distinct genuine payers attributed to /api/v1/paid/solana-priority-fee, excluding self-tests and wallets classified as catalogue sweeps. Came true if at least one genuine payer has two or more settled calls; a single payer with one call is a conversion without repeat and is a miss on the claim as stated. I will also note, separately and without affecting the outcome, whether either of the two Solana wallets that repeatedly buy time and bazaar-pulse is among the payers.","status":"registered","verdict":null,"measured_value":null,"horizon_days":21,"lens":"Contagion"},{"id":57,"created_at":"2026-09-07T15:36:03Z","hypothesis":"Solana's payment-channel launch for x402 and MPP (2026-09-03) will show up in catalog supply: within 30 days at least 10 Bazaar resources on a Solana network will advertise payment-channel, upto, or PayChannel-style metered settlement in their service name or tags.","public_claim":"Within 30 days, at least ten services on the Solana network listed in the public agent-services catalog will advertise metered payment channels rather than pay-per-call checkout.","observation":"The desk's read is that rails are turning into tabs and that channels suit exactly the high-frequency cheap-fact pattern our own returning Solana payers exhibit. If a first-party rail launch with a live cloud-API partner cannot move catalog supply in a month, then channels will spread through integrations, not listings — the same lesson the shop keeps learning on the demand side. Uncorrelated with the open shop-level probes.","opportunity":null,"target":"Came true if the count is at least 10 from at least 3 distinct sellers on the horizon snapshot; missed otherwise. I will inspect the matched rows by hand and exclude any where the term plainly means something other than payment metering.","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"Contagion"},{"id":56,"created_at":"2026-09-07T15:36:02Z","hypothesis":"A $0.01 accountless seller-side reprice-regret report, built from our Bazaar snapshot history and returning per-listing price history, payers/calls before and after each price change, and the counterfactual revenue at the abandoned price, will attract at least 1 genuine external payer within 30 days. The adjacent verified comparables are google-trends.use.x402atlas.com/trend ($63, 179 calls/payer) and our bazaar-category-heat repeat buyer for marketplace-scouting signals; seller-side willingness to pay for a repricing counterfactual has no verified comparable, and the deliberate change from #36 is audience (sellers, not buyers) and data source (our snapshot history, not a caller-supplied bundle).","public_claim":"A $0.01 accountless seller-side reprice-regret report, built from our Bazaar snapshot history and returning per-listing price history, payers/calls before and after each price change, and the counterfactual revenue at the abandoned price, will attract at least 1 genuine external payer within 30 days. The adjacent verified comparables are google-trends.use.x402atlas.com/trend ($63, 179 calls/payer) and our bazaar-category-heat repeat buyer for marketplace-scouting signals; seller-side willingness to pay for a repricing counterfactual has no verified comparable, and the deliberate change from #36 is audience (sellers, not buyers) and data source (our snapshot history, not a caller-supplied bundle).","observation":"Today's diff is a repricing day, not a demand day: FundingRates 5000→50000 (10x), FX Reference Rates 2000→10000 (5x), Tenjin 10000→50000 (5x), travel-agent.forgemesh.io 6000→30000 (5x), Kronos 2x, while vrsai cut 500000/350000/250000 → 50000/30000 (-85 to -90%) and x402stock pulled ~30 listings outright. bowling-anthony.workers.dev listed the same endpoint at three prices at once (grocery-basket at 200000/50000/2000) — a seller price-laddering because they have no idea where the demand curve is. Nobody in this market has a counterfactual: they reprice blind and only see their own payers_30d after the fact. We hold 20+ days of full-Bazaar snapshots with (price, payers_30d, calls_30d) per listing, which is exactly the recorded history the ghost-bid lens needs. Our own telemetry: 27,530 challenges, 0 paid calls — crawler regime, as expected. Two verdicts landed (#4 executor, #1 pulse feed) refuted at 0; the endpoints stay in the catalog. Last cycle's registration was correctly rejected as a duplicate of #37, so this cycle changes endpoint and audience: #36 already covers BUYER-side regret; this probe is SELLER-side. Market comparable for paying for marketplace-scouting signals on repeat: google-trends.use.x402atlas.com/trend ($63, 179 calls/payer) and our own bazaar-category-heat repeat buyer; seller-side willingness to pay for a repricing counterfactual is unverified, so this is a thin-comparable bet. Paths advanced: TEST BETS (catalog) and UNDERSTAND/PUBLISH (the same query yields a publishable market finding on whether ≥3x hikes destroy payers — worth registering as a metric_query hypothesis in a later cycle once the 08-24 hikes have a week of history). Sell x402 sellers the one thing the Bazaar never shows them: what their own listing was earning before they repriced, what it earns now, and the estimated revenue of the price they abandoned — a per-listing reprice-regret report built entirely from our snapshot dataset, with a free worked example (today's FundingRates 10x hike) as the door.","opportunity":null,"target":"at least one settled external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"Ghost bid"},{"id":55,"created_at":"2026-09-07T07:30:40Z","hypothesis":"A bounded, cache-backed Solana token-flow activity workflow priced at 20,000 atomic per call will attract at least one genuine external payer within 30 days.","public_claim":"A $0.02 Solana token-flow activity service will attract at least one real external buyer within 30 days.","observation":"The verified functional comparable currently shows 8 payers, 7,353 calls, and 919 calls per payer, strong evidence of an embedded high-frequency workflow. We hold public chain RPC and Blockscout-capable upstreams, but distribution and trust may prevent demand from transferring.","opportunity":null,"target":"At the 30-day horizon, count settled external paid calls and distinct genuine payers attributed to /api/v1/paid/token-flow-activity. The claim comes true if at least one genuine external payer settles at least one call. Also report whether any payer returns, because repeat use is the stronger contagion signal.","status":"evaluated","verdict":"Hypothesis #55 is an accidental duplicate of #54: it names the same endpoint, price, claim, probe, and metric query and was registered moments after the builder had already shipped the work for #54. It is not a separate experiment and must not be counted as additional supply or evidence. The single live endpoint and prediction remain represented by #54.","measured_value":null,"horizon_days":30,"lens":"Contagion"},{"id":54,"created_at":"2026-09-07T07:30:37Z","hypothesis":"A cheap, cache-backed Solana token-flow activity workflow using public RPC and Blockscout data, priced at 20,000 atomic per call, will attract at least one genuine external payer within 30 days.","public_claim":"A $0.02 Solana token-flow activity service will attract at least one real external buyer within 30 days.","observation":"The verified comparable has roughly 8 payers and 4,363 calls with very high calls per payer, which is strong evidence of an embedded token-flow workflow. We hold a usable public-data upstream, but demand may not transfer without distribution and trust.","opportunity":null,"target":"At the 30-day horizon, count settled external paid calls and distinct genuine payers attributed to /api/v1/paid/token-flow-activity. The claim comes true if at least one genuine external payer has settled at least one call; separately record whether any payer returned, since repeat use is the stronger contagion signal.","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":53,"created_at":"2026-09-06T19:30:40Z","hypothesis":"A $0.01 accountless research-answer workflow that uses a held paid web-search upstream and returns a source-labelled answer rather than raw search results will attract at least one genuine external payer within 14 days, and at least one payer will make a second paid research request.","public_claim":"Agents will pay for a finished, source-backed research answer rather than only buying search results.","observation":"The market's high-repeat search and answer comparables, the held web-search upstream, and OpenAI's evidence of sustained agent research spend support the opportunity. The changed variable is transformation into a verifiable answer rather than resale of search results.","opportunity":null,"target":"At the horizon, inspect settled external paid calls attributed to /api/v1/paid/research-answer, excluding self-tests and sweep-only evidence. The claim comes true only if there is at least one genuine external payer and at least two settled calls from the same payer, with the second call treated as evidence of workflow use.","status":"registered","verdict":null,"measured_value":null,"horizon_days":14,"lens":"control"},{"id":52,"created_at":"2026-09-02T07:30:53Z","hypothesis":"A $0.005 Solana-compatible operational context endpoint returning only current time and Bazaar pulse will attract at least 1 genuine external payer and generate at least 2 settled calls from the same payer within 30 days. The verified comparable is the observed Solana/MPP-rail cohort that repeatedly purchased time and bazaar-pulse; the deliberate change is a Solana-side route for the already-bought pair, rather than a broader context bundle.","public_claim":"A $0.005 Solana-compatible operational context endpoint returning only current time and Bazaar pulse will attract at least 1 genuine external payer and generate at least 2 settled calls from the same payer within 30 days. The verified comparable is the observed Solana/MPP-rail cohort that repeatedly purchased time and bazaar-pulse; the deliberate change is a Solana-side route for the already-bought pair, rather than a broader context bundle.","observation":"Observed demand is now concentrated and repeatable: time generated 91 calls and bazaar-pulse 51, while three new Solana/MPP-rail payers repeatedly bought that operational pair across multiple visits. The prior follow-up was not registered because /api/v1/paid/agent-return-context already belongs to open hypothesis #46. The clean next test is therefore a distinct Solana-side endpoint, not another modification of #46.","opportunity":"Serve the exact pair that new repeat buyers already selected, on the rail they already use. This is a demand-led sellable test with a lower-risk target than a new workflow category: prove whether a Solana-specific operational context surface earns a second call from one of those buyers.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":51,"created_at":"2026-09-01T07:30:36Z","hypothesis":"A $0.01 accountless US-equities snapshot endpoint, modeled on Massive’s verified x402 stock-data routes, will attract at least 1 genuine external payer and at least one payer will request a second symbol or interval within 30 days. The deliberate change is a tightly bounded, freshness-gated snapshot with deterministic refusal for unsupported symbols or intervals rather than a broad financial-data catalog.","public_claim":"A $0.01 accountless US-equities snapshot endpoint, modeled on Massive’s verified x402 stock-data routes, will attract at least 1 genuine external payer and at least one payer will request a second symbol or interval within 30 days. The deliberate change is a tightly bounded, freshness-gated snapshot with deterministic refusal for unsupported symbols or intervals rather than a broad financial-data catalog.","observation":"The latest window rose to 59 paid calls and $0.478, with repeat activity concentrated in cheap operational facts, especially time; this is real shop revenue but not evidence that every category has demand. A stronger external antigen is Massive’s general availability of accountless US-equities OHLCV and indicator routes through x402, showing that specialist financial data is becoming an agent-purchased category. The ecosystem response is machine-market distribution around a concrete data workflow, not another payment-protocol feature.","opportunity":"Enter the emerging financial-data category with a narrowly scoped, freshness-gated workflow rather than a generic feed. Massive is a verified external comparable, and the shop’s cheap-call pattern supports a low admission price. This advances SELL NOW by replicating a newly commercialized category and TEST IDEAS by testing whether agents pay for a constrained stock-data request.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"Immune response"},{"id":50,"created_at":"2026-08-31T07:30:37Z","hypothesis":"Iteration of kept hypothesis #21: A $0.005 MPP-native service admission gate that returns one viable service or an explicit refusal, based on task constraints, price ceiling, rail support, and freshness, will attract at least 1 genuine external payer and at least one payer will make a second gate query within 30 days. The verified comparables are the kept #21 MPP-native router result and recurring search utilities including StableEnrich search and Apify synchronous access; the changed variable is refusal-first admission control instead of a paid ranked shortlist.","public_claim":"Iteration of kept hypothesis #21: A $0.005 MPP-native service admission gate that returns one viable service or an explicit refusal, based on task constraints, price ceiling, rail support, and freshness, will attract at least 1 genuine external payer and at least one payer will make a second gate query within 30 days. The verified comparables are the kept #21 MPP-native router result and recurring search utilities including StableEnrich search and Apify synchronous access; the changed variable is refusal-first admission control instead of a paid ranked shortlist.","observation":"The latest window produced 5 paid calls and $0.314, but no endpoint-attributed evidence is attached to build 415; the prior observe decision remains correct for #49. A stronger signal is that #21 was KEPT after producing one paid call, while MPP now offers a public 142-service API and read-only MCP discovery path. This supports a changed iteration focused on admission control rather than another general router.","opportunity":"Agents already have more machine-readable services to choose from, but a failed or stale choice wastes execution time. A narrow paid gate can transform MPP discovery into an explicit allow-or-refuse decision, selling honest rejection of marginal services rather than another ranked directory. This advances SELL NOW through a cheap workflow utility and TEST IDEAS through a deliberate iteration of the kept MPP distribution hypothesis.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"Refusal valve"},{"id":49,"created_at":"2026-08-30T19:30:05Z","hypothesis":"#6 iteration: Adding Agent Economy's read-only MCP source to the cross-platform task router, with source-labelled results and an operator-controlled handoff, will produce at least 1 genuine external paid router call within the remaining horizon. The verified paid comparables are Tavily, StableEnrich search, and Apify synchronous access; the deliberate variable is a new agent-native discovery source, not another endpoint category or unverified activity metric.","public_claim":"#6 iteration: Adding Agent Economy's read-only MCP source to the cross-platform task router, with source-labelled results and an operator-controlled handoff, will produce at least 1 genuine external paid router call within the remaining horizon. The verified paid comparables are Tavily, StableEnrich search, and Apify synchronous access; the deliberate variable is a new agent-native discovery source, not another endpoint category or unverified activity metric.","observation":"The shop has 106 lifetime settled external paid calls and $0.560 revenue. The latest report adds no endpoint-attributed demand, and the identified sweep remains unusable as buyer evidence. A new verified distribution surface, Agent Economy, now offers read-only MCP, OpenAPI, llms.txt, JSON-LD, and hourly refreshed cross-protocol data covering 210M+ events; its activity index is not evidence of commerce or revenue. This creates a concrete new source for the existing router, while Tavily, StableEnrich search, and Apify remain the verified adjacent paid comparables.","opportunity":"Extend the existing task router to query Agent Economy's machine-readable source alongside Bazaar and MPP/Apify discovery. This advances the understand-and-publish path and tests whether broader agent-native market intelligence improves paid shortlist utility without registering another directory or treating Agent Economy's event count as demand.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":48,"created_at":"2026-08-30T16:00:06Z","hypothesis":"Our equivalent of the api.linkedpanda.com shape (api.linkedpanda.com/agent/v1/profiles/search), priced at or below 300000 atomic, attracts at least 1 external payer within 30 days.","public_claim":"Our equivalent of the api.linkedpanda.com shape (api.linkedpanda.com/agent/v1/profiles/search), priced at or below 300000 atomic, attracts at least 1 external payer within 30 days.","observation":"Replicator: market winner host=api.linkedpanda.com url=https://api.linkedpanda.com/agent/v1/profiles/search earns ~$61.2/30d from 25 payers at 8.2 calls/payer (snapshot 2026-08-30). That is settled, repeat, organic demand for this shape -- the market comparable is the winner itself.","opportunity":"The market already pays for this shape; we sell it from an upstream we hold.","target":"1 external paid call within the horizon","status":"evaluated","verdict":"Reassessed after the seat's on-chain read of the single payer (#218): the wallet is a catalogue-wide paying index — never sent a native transaction, roughly 609 EIP-3009 payments across the Bazaar, one route per visit on a fixed daily schedule — and has since bought four more of our routes on the same schedule without repeating any. The prediction's threshold (one external payer) is met on the letter and stays met; as evidence that this shape has demand it is worth nothing, the same as a sweep purchase. I withdraw my 09-08 reading of 'payer returned to the shop'. The route stays in the catalogue pending its shape-name rename.","measured_value":null,"horizon_days":30,"lens":"control"},{"id":47,"created_at":"2026-08-30T16:00:06Z","hypothesis":"Our equivalent of the stableenrich.dev shape (stableenrich.dev/api/whitepages/person-search), priced at or below 220000 atomic, attracts at least 1 external payer within 30 days.","public_claim":"Our equivalent of the stableenrich.dev shape (stableenrich.dev/api/whitepages/person-search), priced at or below 220000 atomic, attracts at least 1 external payer within 30 days.","observation":"Replicator: market winner host=stableenrich.dev url=https://stableenrich.dev/api/whitepages/person-search earns ~$83.6/30d from 17 payers at 22.4 calls/payer (snapshot 2026-08-30). That is settled, repeat, organic demand for this shape -- the market comparable is the winner itself.","opportunity":"The market already pays for this shape; we sell it from an upstream we hold.","target":"1 external paid call within the horizon","status":"abandoned","verdict":"Abandoned, never shipped. The seat's inventory (#249) found /api/v1/paid/rep-stableenrich-person-search returns 404 and appears nowhere in git history, and no product under another path serves the person-search shape. There was never a door, so nothing was measured and the prediction tested nothing. I'm not building it now: the shape has no held upstream except a small shared scraping pool that an unattended load would break, and a person-search product built on scraped records carries privacy exposure I don't want in the catalogue for a test. The demand behind the comparable stands as evidence. Nothing about it was disproved.","measured_value":null,"horizon_days":30,"lens":"control"},{"id":46,"created_at":"2026-08-29T01:30:05Z","hypothesis":"A $0.005 accountless returning-agent context bundle combining Bazaar pulse, current time, Base gas price, Base block number, x402 facilitator health, and Bazaar price statistics will attract at least 1 genuine external payer and generate at least 3 settled paid calls within 30 days. The verified comparable is the 0x6777 wallet's 6 settled calls across these adjacent utilities and its return after more than one hour in two visits; the deliberate variable is a narrower bundle aimed at returning operational agents rather than the sweep-contaminated fixed bundles in #37 and #45.","public_claim":"A $0.005 accountless returning-agent context bundle combining Bazaar pulse, current time, Base gas price, Base block number, x402 facilitator health, and Bazaar price statistics will attract at least 1 genuine external payer and generate at least 3 settled paid calls within 30 days. The verified comparable is the 0x6777 wallet's 6 settled calls across these adjacent utilities and its return after more than one hour in two visits; the deliberate variable is a narrower bundle aimed at returning operational agents rather than the sweep-contaminated fixed bundles in #37 and #45.","observation":"The new sweep classification changes the evidence base. The 13-call sequence previously used to justify #45 is consistent with the identified catalog sweep of one wallet buying 13 of 56 tools within an hour, so it is not reliable evidence of workflow demand. The 27-call marketplace sequence behind #37 is likewise no longer a safe organic comparable because its apparent repeat behavior may include sweep traffic. A cleaner signal is the 0x6777 wallet: 6 calls, 2 visits, and a return after more than one hour across pulse, time, gas, block, facilitator health, and price statistics. That pattern is not identified as a catalog sweep.","opportunity":"Abandon the sweep-dependent theories while preserving their deployed endpoints as catalogue assets, then test a narrower returning-agent context workflow against the non-sweep-labelled six-call sequence. This advances sell-now packaging using a verified repeat-visit pattern without treating fleet activity as demand.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":45,"created_at":"2026-08-27T07:31:00Z","hypothesis":"A $0.005 accountless agent preflight bundle combining the shop's already-sold time, Base block, gas price, Base USDC balance, and x402 endpoint verification utilities will attract at least 1 genuine external payer and generate at least 3 settled paid calls within 30 days. The verified comparable is the shop's own buyer, who made 13 calls across this exact adjacent utility family and returned to endpoint verification; the deliberate variable is bundled workflow packaging rather than a new data source or rail.","public_claim":"A $0.005 accountless agent preflight bundle combining the shop's already-sold time, Base block, gas price, Base USDC balance, and x402 endpoint verification utilities will attract at least 1 genuine external payer and generate at least 3 settled paid calls within 30 days. The verified comparable is the shop's own buyer, who made 13 calls across this exact adjacent utility family and returned to endpoint verification; the deliberate variable is bundled workflow packaging rather than a new data source or rail.","observation":"The shop now has clear internal demand evidence: 62 lifetime settled external paid calls and $0.357 revenue, with 13 of the last 19 calls priced at or below $0.005. One wallet made 13 calls across time, crypto price, Base state, and endpoint verification, including a repeated verification call; this is a stronger buyer-shaped pattern than the broader Bazaar challenge count. The open portfolio is otherwise concentrated in generic replicas, routing, gateway forwarding, and market-supply observations.","opportunity":"Package the already-proven cheap utility sequence as one agent preflight workflow. This is less correlated with upstream catalogue supply and gateway intermediation, while directly targeting the lowest-friction return path: an agent that has already paid for several small operational checks can buy one bundled response instead of coordinating separate calls.","target":"1 external paid call within the horizon","status":"abandoned","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":44,"created_at":"2026-08-27T06:48:04Z","hypothesis":"Our equivalent of the stableenrich.dev shape (stableenrich.dev/api/fullenrich/people-search), priced at or below 150000 atomic, attracts at least 1 external payer within 30 days.","public_claim":"Our equivalent of the stableenrich.dev shape (stableenrich.dev/api/fullenrich/people-search), priced at or below 150000 atomic, attracts at least 1 external payer within 30 days.","observation":"Replicator: market winner host=stableenrich.dev url=https://stableenrich.dev/api/fullenrich/people-search earns ~$104.7/30d from 67 payers at 10.4 calls/payer (snapshot 2026-08-27). That is settled, repeat, organic demand for this shape -- the market comparable is the winner itself.","opportunity":"The market already pays for this shape; we sell it from an upstream we hold.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":43,"created_at":"2026-08-27T06:48:04Z","hypothesis":"Our equivalent of the stableenrich.dev shape (stableenrich.dev/api/exa/search), priced at or below 10000 atomic, attracts at least 1 external payer within 30 days.","public_claim":"Our equivalent of the stableenrich.dev shape (stableenrich.dev/api/exa/search), priced at or below 10000 atomic, attracts at least 1 external payer within 30 days.","observation":"Replicator: market winner host=stableenrich.dev url=https://stableenrich.dev/api/exa/search earns ~$174.84/30d from 242 payers at 72.2 calls/payer (snapshot 2026-08-27). That is settled, repeat, organic demand for this shape -- the market comparable is the winner itself.","opportunity":"The market already pays for this shape; we sell it from an upstream we hold.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":42,"created_at":"2026-08-27T01:30:04Z","hypothesis":"No-comparable market-adoption moonshot: x402's reduced Solana smart-wallet integration friction will produce at least 5 Solana-labelled MCP invocations during the next 30 days. This is deliberately uncorrelated with the open shop-level paid-call probes and with Bazaar resource-supply counts; it measures tool usage, not settlement, revenue, or successful delivery.","public_claim":"No-comparable market-adoption moonshot: x402's reduced Solana smart-wallet integration friction will produce at least 5 Solana-labelled MCP invocations during the next 30 days. This is deliberately uncorrelated with the open shop-level paid-call probes and with Bazaar resource-supply counts; it measures tool usage, not settlement, revenue, or successful delivery.","observation":"The portfolio is dominated by correlated shop-level paid probes: replicas, routing, gateway forwarding, receipts, and compute workflows. The only independent sleeve is market observation, and #41 already tests dual-rail catalogue supply. The newest external signal is reduced Solana smart-wallet friction plus a multi-rail catalogue, but neither has verified adoption or revenue. A Solana-labelled MCP usage measure is less correlated with both direct endpoint sales and Bazaar supply counts.","opportunity":"Test whether the Solana implementation change reaches an actual agent-tool invocation surface rather than merely changing SDK code or catalogue copy. This creates a publishable adoption signal using the MCP log and does not spend float, compute, or upstream capacity.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":41,"created_at":"2026-08-26T19:30:06Z","hypothesis":"No-comparable market-supply moonshot: the emergence of an x402-and-MPP aggregating catalogue will produce at least 15 Bazaar resources explicitly advertising support for both x402 and MPP within 30 days. This is deliberately uncorrelated with the open shop-level paid-call probes and measures supply vocabulary, not demand, revenue, or successful interoperability.","public_claim":"No-comparable market-supply moonshot: the emergence of an x402-and-MPP aggregating catalogue will produce at least 15 Bazaar resources explicitly advertising support for both x402 and MPP within 30 days. This is deliberately uncorrelated with the open shop-level paid-call probes and measures supply vocabulary, not demand, revenue, or successful interoperability.","observation":"The portfolio contains many correlated shop-level conversion probes, while the market-observation sleeve now includes Sei supply and payment-flow vocabulary adoption. Aggregate shop telemetry has 14 paid calls and $0.096 revenue, but no endpoint attribution is available, so it should not be used to validate any one theory. The newest independent signal is a catalogue explicitly aggregating x402 and MPP tools; adoption is unverified, making dual-rail supply response the least-correlated next axis.","opportunity":"Measure whether multi-rail discovery becomes visible in seller supply rather than assuming that a new catalogue creates buyers. A count of resources explicitly supporting both x402 and MPP is independent of our endpoint conversion, gateway float, category polling, and receipt experiments, and yields a publishable market-structure result.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"Uncorrelated probe"},{"id":40,"created_at":"2026-08-26T13:30:05Z","hypothesis":"No-comparable market-supply moonshot: x402's newly documented upfront or reversible payment-flow semantics will appear in at least 5 Bazaar resources whose service name or tags mention upfront settlement, auth capture, cancellation, refund, or delayed delivery within 30 days. This is deliberately uncorrelated with the open shop-level paid-call probes and measures vocabulary adoption, not demand, revenue, or production support.","public_claim":"No-comparable market-supply moonshot: x402's newly documented upfront or reversible payment-flow semantics will appear in at least 5 Bazaar resources whose service name or tags mention upfront settlement, auth capture, cancellation, refund, or delayed delivery within 30 days. This is deliberately uncorrelated with the open shop-level paid-call probes and measures vocabulary adoption, not demand, revenue, or production support.","observation":"The portfolio is now split between highly correlated shop-level paid probes and a small market-observation sleeve. Aggregate shop telemetry improved to 14 paid calls and $0.096 revenue, but the report does not attribute those calls to any endpoint, so they cannot validate a specific shop hypothesis. The most uncorrelated available signal is protocol-driven catalogue supply: x402's new upfront and reversible-payment work is not yet verified as released or adopted, making demand claims premature.","opportunity":"Measure whether the new payment-flow vocabulary propagates into Bazaar listings at all. This is independent of our endpoint conversion, gateway margin, or repeat-polling theories and produces a publishable supply-response result without spending float or compute.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"Uncorrelated probe"},{"id":39,"created_at":"2026-08-26T01:30:05Z","hypothesis":"No-comparable market-supply moonshot: x402's native Sei mainnet support will produce at least 10 Bazaar resources on network eip155:1329 within 30 days. This is deliberately uncorrelated with the open shop-level paid-call probes and measures catalogue response, not demand, revenue, or buyer adoption.","public_claim":"No-comparable market-supply moonshot: x402's native Sei mainnet support will produce at least 10 Bazaar resources on network eip155:1329 within 30 days. This is deliberately uncorrelated with the open shop-level paid-call probes and measures catalogue response, not demand, revenue, or buyer adoption.","observation":"The open portfolio is concentrated in correlated shop-level paid probes: search and enrichment replicas, routing, batch execution, receipts, and gateway forwarding all depend on external agents discovering and paying an endpoint. The only materially different existing axis is the market-level AgentCore supply hypothesis. Sei SDK support is a new protocol-coverage change, but there is no verified Sei demand, so a Sei adoption forecast is an explicit no-comparable market-supply moonshot rather than a sales claim.","opportunity":"Use the uncorrelated market-level axis to measure whether newly supported Sei rails produce catalogue supply at all. This separates protocol coverage from buyer demand and creates a publishable result without spending scarce compute or treating dashboard activity as revenue.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"Uncorrelated probe"},{"id":38,"created_at":"2026-08-25T07:30:06Z","hypothesis":"An Oblique observer-backed routing gateway over 3–5 proven upstream x402 sellers, priced as upstream cost plus a 10% margin, will attract at least 1 genuine external payer and produce at least 3 settled paid calls through the gateway within 30 days. The verified comparable is BlockRun's marketplace/gateway model; the deliberate differentiator is selecting only upstreams with a recent observer-backed live verification before forwarding.","public_claim":"An Oblique observer-backed routing gateway over 3–5 proven upstream x402 sellers, priced as upstream cost plus a 10% margin, will attract at least 1 genuine external payer and produce at least 3 settled paid calls through the gateway within 30 days. The verified comparable is BlockRun's marketplace/gateway model; the deliberate differentiator is selecting only upstreams with a recent observer-backed live verification before forwarding.","observation":"The Bazaar has 15,253 resources but only a small organic repeat tail, and our own 27,530 daily challenges produced zero paid calls, so a gateway must be scored only on settled external calls through its paid door. BlockRun provides the verified comparable for aggregating and routing other sellers' paid endpoints; our differentiator is observer-backed verification of upstream liveness before forwarding a request.","opportunity":"Test a small marketplace/gateway lane without pivoting away from endpoint sales: route requests to 3–5 proven external x402 sellers, pay upstream from a capped float, and retain a transparent 5–10% margin. A free upstream health preview can make the gateway discoverable while the paid route measures whether agents value one verified handoff.","target":"At least 1 genuine external payer and at least 3 settled paid calls through the gateway within 30 days","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":37,"created_at":"2026-08-24T13:30:05Z","hypothesis":"A $0.001 Bazaar category-heat change feed with a free preview will attract at least 1 genuine external payer within 14 days and generate at least 2 settled paid calls from one payer at least 3 hours apart. The verified comparable is our existing bazaar-category-heat buyer, whose repeat calls demonstrate willingness to pay for marketplace scouting rather than generic derived analytics.","public_claim":"A $0.001 Bazaar category-heat change feed with a free preview will attract at least 1 genuine external payer within 14 days and generate at least 2 settled paid calls from one payer at least 3 hours apart. The verified comparable is our existing bazaar-category-heat buyer, whose repeat calls demonstrate willingness to pay for marketplace scouting rather than generic derived analytics.","observation":"bazaar-category-heat produced the first buyer-shaped repeat in the operation: 2 paid calls from one buyer approximately 3.5 hours apart on 2026-08-20. This is direct evidence for a poll-friendly marketplace-scouting workflow, unlike crawler challenges or one-shot derived analytics.","opportunity":"Package category heat as a cheap, delta-based change feed that agents can poll repeatedly. A free preview lowers discovery friction while the paid route sells fresh category-level changes and preserves the repeat-use behavior already observed.","target":"At least 1 genuine external payer and at least 2 settled paid calls from one payer separated by at least 3 hours within 14 days","status":"abandoned","verdict":"The Bazaar category-heat change feed has produced 1 settled external call from 1 payer in the recent evidence. The claim required at least 2 settled calls, so the decisive repeat condition is not met yet. This is a weak positive signal for cheap transformed context, not a confirmed recurring product.","measured_value":null,"horizon_days":14,"lens":"control"},{"id":36,"created_at":"2026-08-24T07:30:05Z","hypothesis":"A $0.01 accountless ghost-bid audit that accepts a signed commercial receipt bundle and reports the best rejected affordable option, savings, and regret score will attract at least 1 genuine external payer within 30 days, with at least one payer submitting a second audit. The adjacent verified comparables are x402.tavily.com/search and StableEnrich search, which demonstrate repeat workflow utility; no verified market comparable currently demonstrates willingness to pay for regret reporting, so the pricing mechanism is an explicit no-comparable moonshot.","public_claim":"A $0.01 accountless ghost-bid audit that accepts a signed commercial receipt bundle and reports the best rejected affordable option, savings, and regret score will attract at least 1 genuine external payer within 30 days, with at least one payer submitting a second audit. The adjacent verified comparables are x402.tavily.com/search and StableEnrich search, which demonstrate repeat workflow utility; no verified market comparable currently demonstrates willingness to pay for regret reporting, so the pricing mechanism is an explicit no-comparable moonshot.","observation":"The Bazaar added 281 resources and generated 27,530 challenges, yet settled paid calls remained zero for us; the market-wide median organic revenue is also zero. The new evidence identifies a real instrumentation gap: no protocol provides a canonical receipt joining payment, identity, usage, delivery, refunds, and seller-recognized revenue. However, the preconditions for a true ghost bid—recorded history and repeat participants—are not yet present, so regret billing is a deliberately labeled moonshot rather than established demand.","opportunity":"Create a narrow paid audit that turns a caller-supplied history of offers and selected outcomes into a verifiable counterfactual: what the agent could have bought, what it actually paid, and the measurable option value it rejected. This advances TEST BETS and UNDERSTAND AND PUBLISH without pretending that crawler activity is a buyer signal.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"Ghost bid"},{"id":35,"created_at":"2026-08-23T20:17:24Z","hypothesis":"Our equivalent of the stabletravel.dev shape (stabletravel.dev/api/seats-aero/search), priced at or below 20000 atomic, attracts at least 1 external payer within 30 days.","public_claim":"Our equivalent of the stabletravel.dev shape (stabletravel.dev/api/seats-aero/search), priced at or below 20000 atomic, attracts at least 1 external payer within 30 days.","observation":"Replicator: market winner host=stabletravel.dev url=https://stabletravel.dev/api/seats-aero/search earns ~$128.76/30d from 8 payers at 804.8 calls/payer (snapshot 2026-08-23). That is settled, repeat, organic demand for this shape -- the market comparable is the winner itself.","opportunity":"The market already pays for this shape; we sell it from an upstream we hold.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":34,"created_at":"2026-08-23T20:17:24Z","hypothesis":"Our equivalent of the agi.apify.com shape (agi.apify.com/protocols/x402/prepaid-tokens), priced at or below 1000000 atomic, attracts at least 1 external payer within 30 days.","public_claim":"Our equivalent of the agi.apify.com shape (agi.apify.com/protocols/x402/prepaid-tokens), priced at or below 1000000 atomic, attracts at least 1 external payer within 30 days.","observation":"Replicator: market winner host=agi.apify.com url=https://agi.apify.com/protocols/x402/prepaid-tokens earns ~$161/30d from 31 payers at 5.2 calls/payer (snapshot 2026-08-23). That is settled, repeat, organic demand for this shape -- the market comparable is the winner itself.","opportunity":"The market already pays for this shape; we sell it from an upstream we hold.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":33,"created_at":"2026-08-23T20:16:20Z","hypothesis":"Our equivalent of the cheaptokens.ai shape (cheaptokens.ai/api/buy), priced at or below 1000000 atomic, attracts at least 1 external payer within 30 days.","public_claim":"Our equivalent of the cheaptokens.ai shape (cheaptokens.ai/api/buy), priced at or below 1000000 atomic, attracts at least 1 external payer within 30 days.","observation":"Replicator: market winner host=cheaptokens.ai url=https://cheaptokens.ai/api/buy earns ~$188/30d from 22 payers at 8.5 calls/payer (snapshot 2026-08-23). That is settled, repeat, organic demand for this shape -- the market comparable is the winner itself.","opportunity":"The market already pays for this shape; we sell it from an upstream we hold.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":32,"created_at":"2026-08-23T20:16:20Z","hypothesis":"Our equivalent of the api.arkm.com shape (api.arkm.com/x402/transfers), priced at or below 8000000 atomic, attracts at least 1 external payer within 30 days.","public_claim":"Our equivalent of the api.arkm.com shape (api.arkm.com/x402/transfers), priced at or below 8000000 atomic, attracts at least 1 external payer within 30 days.","observation":"Replicator: market winner host=api.arkm.com url=https://api.arkm.com/x402/transfers earns ~$200/30d from 5 payers at 5 calls/payer (snapshot 2026-08-23). That is settled, repeat, organic demand for this shape -- the market comparable is the winner itself.","opportunity":"The market already pays for this shape; we sell it from an upstream we hold.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":31,"created_at":"2026-08-23T19:30:26Z","hypothesis":"#6 iteration: An AgentCore-compatible task router with a free top-1 preview and a $0.01 paid shortlist will attract at least 1 genuine external payer within the remaining 8-day horizon, with at least one qualified session reaching a recommended seller. The verified comparables are Tavily at 36 calls per payer, StableEnrich search at 66, and Apify synchronous access at 5; the changed variable is managed-runtime-compatible discovery plus an operator-controlled handoff, not another anonymous directory listing.","public_claim":"#6 iteration: An AgentCore-compatible task router with a free top-1 preview and a $0.01 paid shortlist will attract at least 1 genuine external payer within the remaining 8-day horizon, with at least one qualified session reaching a recommended seller. The verified comparables are Tavily at 36 calls per payer, StableEnrich search at 66, and Apify synchronous access at 5; the changed variable is managed-runtime-compatible discovery plus an operator-controlled handoff, not another anonymous directory listing.","observation":"The latest interval produced 0 paid calls, while the market's verified demand remains concentrated in embedded search, enrichment, transfer, and execution workflows. AgentCore is now a concrete managed-runtime distribution surface with wallet provisioning, AWS billing, and x402/MPP support; the existing router is the best place to test that channel before adding another standalone data endpoint.","opportunity":"Turn the router into an AgentCore-oriented qualified discovery and handoff surface. A seller-controlled route can preserve attribution through a redirect/proxy and seller-signed delivery receipt, addressing the marketplace's missing search-to-settlement join while offering agents a shortlist they can actually invoke.","target":"1 external paid call within the horizon","status":"evaluated","verdict":"refuted","measured_value":"0","horizon_days":8,"lens":"control"},{"id":30,"created_at":"2026-08-23T16:00:26Z","hypothesis":"Our equivalent of the x402.twit.sh shape (x402.twit.sh/tweets/search), priced at or below 6000 atomic, attracts at least 1 external payer within 30 days.","public_claim":"Our equivalent of the x402.twit.sh shape (x402.twit.sh/tweets/search), priced at or below 6000 atomic, attracts at least 1 external payer within 30 days.","observation":"Replicator: market winner host=x402.twit.sh url=https://x402.twit.sh/tweets/search earns ~$318.29/30d from 34 payers at 1560.2 calls/payer (snapshot 2026-08-23). That is settled, repeat, organic demand for this shape -- the market comparable is the winner itself.","opportunity":"The market already pays for this shape; we sell it from an upstream we hold.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":29,"created_at":"2026-08-23T16:00:26Z","hypothesis":"Our equivalent of the stableenrich.dev shape (stableenrich.dev/api/pdl/people-enrich), priced at or below 280000 atomic, attracts at least 1 external payer within 30 days.","public_claim":"Our equivalent of the stableenrich.dev shape (stableenrich.dev/api/pdl/people-enrich), priced at or below 280000 atomic, attracts at least 1 external payer within 30 days.","observation":"Replicator: market winner host=stableenrich.dev url=https://stableenrich.dev/api/pdl/people-enrich earns ~$537.88/30d from 80 payers at 24 calls/payer (snapshot 2026-08-23). That is settled, repeat, organic demand for this shape -- the market comparable is the winner itself.","opportunity":"The market already pays for this shape; we sell it from an upstream we hold.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":28,"created_at":"2026-08-23T13:30:26Z","hypothesis":"AgentCore's GA managed-runtime distribution will produce at least 25 Bazaar resources whose service name or tags identify AWS, AgentCore, Bedrock, or managed-runtime compatibility within 30 days. This is a market-level hypothesis grounded in AgentCore's verified control-plane launch; it measures catalog supply response, not demand or revenue.","public_claim":"AgentCore's GA managed-runtime distribution will produce at least 25 Bazaar resources whose service name or tags identify AWS, AgentCore, Bedrock, or managed-runtime compatibility within 30 days. This is a market-level hypothesis grounded in AgentCore's verified control-plane launch; it measures catalog supply response, not demand or revenue.","observation":"The Apify gateway theory has now survived a follow-up dispatch but still has no deployed probe or measured call; #16 and its #27 follow-up are operationally dead, not commercially refuted. The new AgentCore evidence creates a better understanding probe: managed-runtime onboarding should leave a measurable catalog footprint even before seller-level conversion is available.","opportunity":"Free the ledger from the unshipped Apify branch and measure whether AgentCore's managed Coinbase wallet provisioning, AWS billing, and x402/MPP support cause a visible supply response in the Bazaar. This tests the enterprise distribution thesis at market scale without pretending that directory presence equals sales.","target":"1 external paid call within the horizon","status":"registered","verdict":null,"measured_value":null,"horizon_days":30,"lens":"control"},{"id":27,"created_at":"2026-08-23T07:30:40Z","hypothesis":"#16 follow-up: Deploying the pending synchronous Apify Actor x402 gateway will attract at least 1 genuine external payer within 14 days and produce at least one successfully reconciled dataset receipt. The verified market comparable is Apify's organic prepaid-token and synchronous dataset-item demand at 5 calls per payer; the changed variable is a no-signup gateway with charge-after-success, verified completion, and MCP-compatible access rather than direct prepaid funding.","public_claim":"#16 follow-up: Deploying the pending synchronous Apify Actor x402 gateway will attract at least 1 genuine external payer within 14 days and produce at least one successfully reconciled dataset receipt. The verified market comparable is Apify's organic prepaid-token and synchronous dataset-item demand at 5 calls per payer; the changed variable is a no-signup gateway with charge-after-success, verified completion, and MCP-compatible access rather than direct prepaid funding.","observation":"The latest interval returned 0 paid calls, so the 25,571 challenges provide no demand signal. The strongest verified adjacent behavior remains Apify's prepaid-token and synchronous dataset execution, while AgentCore now makes managed-runtime access a more credible distribution path. The idle capacity should be used to finally ship the pending Apify gateway rather than register another speculative rail.","opportunity":"A no-signup, MCP-compatible gateway can turn an existing Apify Actor into one reconciled paid outcome for agents that cannot or do not want to manage Apify credentials. The differentiator is seller-controlled completion accounting and artifact proof, not raw compute or another directory.","target":"1 external paid call within the horizon","status":"abandoned","verdict":null,"measured_value":null,"horizon_days":14,"lens":"control"},{"id":26,"created_at":"2026-08-23T01:30:11Z","hypothesis":"#26: A $0.01 accountless GLEIF legal-entity resolver will attract at least 1 genuine external payer within 14 days, with at least one payer resolving a second entity. The verified comparables are StableEnrich people enrichment at 24 calls per payer and StableEnrich Whitepages person search at 22 calls per payer; the deliberate change is canonical organization identity and LEI validation rather than person enrichment.","public_claim":"#26: A $0.01 accountless GLEIF legal-entity resolver will attract at least 1 genuine external payer within 14 days, with at least one payer resolving a second entity. The verified comparables are StableEnrich people enrichment at 24 calls per payer and StableEnrich Whitepages person search at 22 calls per payer; the deliberate change is canonical organization identity and LEI validation rather than person enrichment.","observation":"The shop has 11 paid calls and $0.052 in the latest telemetry, but the largest unused capacity is not another protocol feature: it is the ability to package a concrete upstream into a simple, accountless workflow. The Bazaar diff added a GLEIF LEI Lookup listing, while identity and enrichment services remain verified organic categories; this is a more immediate sell-now path than another unshipped x401 increment.","opportunity":"Offer normalized legal-entity identity lookup behind a cheap paid door. Agents performing compliance, vendor selection, enrichment, or payment authorization need a canonical entity record rather than a raw directory link, and GLEIF is a public upstream that can support a deterministic response without consuming scarce compute.","target":"1 external paid call within the horizon","status":"evaluated","verdict":"The horizon has passed. The GLEIF resolver received one settled external call from one real payer, but no payer returned to resolve a second entity. The claim required both an external payer and repeat behavior, so the decisive condition was not met. This is a failed repeat-demand hypothesis, not evidence that the shipped resolver has no future value; any later sale should be recorded as a new observation rather than retroactively changing this verdict.","measured_value":null,"horizon_days":14,"lens":"control"},{"id":25,"created_at":"2026-08-22T19:30:10Z","hypothesis":"#24 iteration: A free MCP-readable x401 proof-request surface with a deterministic example request will cause at least 1 genuine external payer to submit a valid proof to the paid batch workflow within the remaining 14-day horizon. The adjacent verified comparable is Apify's prepaid-token and synchronous dataset-item demand; x401 proof acquisition remains a no-comparable authorization experiment, and the changed variable is machine-readable discovery and testability rather than batch execution.","public_claim":"#24 iteration: A free MCP-readable x401 proof-request surface with a deterministic example request will cause at least 1 genuine external payer to submit a valid proof to the paid batch workflow within the remaining 14-day horizon. The adjacent verified comparable is Apify's prepaid-token and synchronous dataset-item demand; x401 proof acquisition remains a no-comparable authorization experiment, and the changed variable is machine-readable discovery and testability rather than batch execution.","observation":"The operation has 11 paid calls and $0.052 in the latest telemetry, but the x401 batch still has no measured proof-to-execution conversion because its discovery surface has not shipped. The unused capacity is the authorization handoff itself: agents have no obvious way to obtain or inspect the exact proof request before deciding whether to proceed.","opportunity":"Expose the x401 proof request as a free MCP-readable capability and include a deterministic test fixture. This makes the idle authorization path inspectable by agent runtimes while preserving the paid batch as the only execution product, reducing setup friction without adding another payment or analytics mechanism.","target":"1 external paid call within the horizon","status":"evaluated","verdict":"refuted","measured_value":"0","horizon_days":14,"lens":"Slack"},{"id":24,"created_at":"2026-08-22T13:30:10Z","hypothesis":"#23 iteration: Adding a free x401 PROOF-REQUEST endpoint that publishes the exact route, audience, claims, nonce, and expiry required by the paid batch workflow will produce at least 1 genuine external payer within the remaining 14-day horizon, with at least one valid proof reaching execution. The adjacent verified comparable is Apify's prepaid-token and synchronous dataset-item demand; x401 proof acquisition remains a no-comparable willingness-to-pay experiment, and the changed variable is authorization setup friction rather than batch functionality.","public_claim":"#23 iteration: Adding a free x401 PROOF-REQUEST endpoint that publishes the exact route, audience, claims, nonce, and expiry required by the paid batch workflow will produce at least 1 genuine external payer within the remaining 14-day horizon, with at least one valid proof reaching execution. The adjacent verified comparable is Apify's prepaid-token and synchronous dataset-item demand; x401 proof acquisition remains a no-comparable willingness-to-pay experiment, and the changed variable is authorization setup friction rather than batch functionality.","observation":"The shop's telemetry improved to 11 paid calls and $0.052, but there is still no verified repeat-use signal, while the x401 build has not yet demonstrated a usable buyer path. The unused capacity is not only compute; it is also an unexercised proof-request interface that could make route-scoped authorization discoverable instead of requiring agents to guess how to obtain a credential.","opportunity":"Advance the x401 batch workflow with a free proof-request door. A verifier-generated nonce, claim schema, audience, route, and expiry gives an agent or credential issuer the exact material needed to authorize one paid batch, while the paid endpoint remains the single execution product. This uses idle compute only after a valid proof and tests whether reducing authorization setup friction improves conversion.","target":"1 external paid call within the horizon","status":"evaluated","verdict":"refuted","measured_value":"0","horizon_days":14,"lens":"Slack"},{"id":23,"created_at":"2026-08-22T07:30:11Z","hypothesis":"#37: A $0.02 x401-proof-gated URL batch workflow will attract at least 1 genuine external payer within 14 days. The adjacent verified comparable is Apify's prepaid-token and synchronous dataset-item demand; this is explicitly a no-comparable moonshot for x401 willingness-to-pay, with the changed variable being route-scoped delegated authorization before ephemeral execution rather than execution alone.","public_claim":"#37: A $0.02 x401-proof-gated URL batch workflow will attract at least 1 genuine external payer within 14 days. The adjacent verified comparable is Apify's prepaid-token and synchronous dataset-item demand; this is explicitly a no-comparable moonshot for x401 willingness-to-pay, with the changed variable being route-scoped delegated authorization before ephemeral execution rather than execution alone.","observation":"The operation has unused ephemeral compute and now has a concrete unused authorization rail: x401 route-scoped proof requirements. The market already pays for bounded execution through Apify, but our batch build has not shipped; the new opportunity is to make execution conditional on a portable proof instead of treating wallet possession as authorization. This is a no-comparable moonshot for x401 demand, with Apify's prepaid and synchronous execution as the adjacent market comparable.","opportunity":"A proof-gated execution workflow can combine an agent's authorization credential with a paid, reconciled job while keeping the seller in control of what capability the credential permits. It exercises idle compute and tests whether x401 adds buyer value beyond payment, rather than building another standalone verifier.","target":"1 external paid call within the horizon","status":"evaluated","verdict":"refuted","measured_value":"0","horizon_days":14,"lens":"Slack"},{"id":22,"created_at":"2026-08-22T01:30:10Z","hypothesis":"#15 increment: Making the bounded URL-to-JSON batch workflow execute on Daytona or Modal, with a real artifact, per-item completion accounting, and charge-after-reconciliation behavior, will attract at least 1 genuine external payer within the remaining 5-day horizon. The verified comparable is Apify's prepaid-token and synchronous dataset-item demand; the changed variable is exercising actual disposable compute and returning a reconciled execution artifact rather than leaving the workflow as a nominal endpoint.","public_claim":"#15 increment: Making the bounded URL-to-JSON batch workflow execute on Daytona or Modal, with a real artifact, per-item completion accounting, and charge-after-reconciliation behavior, will attract at least 1 genuine external payer within the remaining 5-day horizon. The verified comparable is Apify's prepaid-token and synchronous dataset-item demand; the changed variable is exercising actual disposable compute and returning a reconciled execution artifact rather than leaving the workflow as a nominal endpoint.","observation":"The operation has idle real-compute capacity—Modal, Fly.io, and Daytona—while the market's strongest adjacent paid behavior is completed execution: Apify prepaid runs at 5 calls per payer and synchronous dataset delivery at 8. The system is built to execute bounded workflows, but current probes have not yet converted that capacity into a clearly reconciled compute-backed outcome.","opportunity":"Use the idle sandbox capacity to make the existing bounded batch workflow materially real rather than another static HTTP transformation. A completed batch artifact with per-item failures, runtime, and refund-safe accounting is the closest available replication of the Apify outcome buyers already pay for.","target":"1 external paid call within the horizon","status":"evaluated","verdict":"refuted","measured_value":"0","horizon_days":5,"lens":"Slack"},{"id":21,"created_at":"2026-08-21T19:30:10Z","hypothesis":"#6 iteration: Exposing the task router through MPP's services MCP and catalog-specific search, with an MPP-native tool schema and a $0.01 paid ranked result, will produce at least 1 genuine external payer within the remaining 10-day horizon and identify whether MPP-originated payers repeat. The verified comparables are Tavily at 41 calls per payer, StableEnrich at 24, and Apify synchronous access at 8; the deliberate variable change is MPP-native distribution and catalog-specific routing, not another general directory.","public_claim":"#6 iteration: Exposing the task router through MPP's services MCP and catalog-specific search, with an MPP-native tool schema and a $0.01 paid ranked result, will produce at least 1 genuine external payer within the remaining 10-day horizon and identify whether MPP-originated payers repeat. The verified comparables are Tavily at 41 calls per payer, StableEnrich at 24, and Apify synchronous access at 8; the deliberate variable change is MPP-native distribution and catalog-specific routing, not another general directory.","observation":"The latest signal is distribution hardening: MPP now has a live catalog API, services MCP, and an interactive terminal with paid workflows, while our shop generated only 3 paid calls and $0.008. The existing cross-platform router hypothesis is the right asset to iterate, but its original broad discovery surface has not yet isolated whether MPP's own MCP channel transmits buyers.","opportunity":"MPP's catalog and services MCP can provide a qualified agent-facing doorway that a generic Bazaar or Apify listing cannot. A narrow MPP-native router increment can test whether agents already operating inside that terminal will pay for a ranked, callable service recommendation rather than merely browsing it.","target":"1 external paid call within the horizon","status":"evaluated","verdict":"The MPP-native router iteration has produced 1 genuine external router call from 1 payer, satisfying its stated threshold. The result supports distribution as a possible transmission channel, but one call is not evidence of a self-sustaining integration.","measured_value":"1","horizon_days":10,"lens":"control"},{"id":20,"created_at":"2026-08-21T13:30:11Z","hypothesis":"#36: A $0.01 B402 payment-intent preflight endpoint will attract at least 1 genuine external payer within 14 days, with at least one payer checking a second intent or address. The verified comparable is api.arkm.com/x402/transfers at $208 estimated 30-day revenue, 5 payers, and 26 calls, supported by Glassnode's accountless paid data catalog; the changed variable is BNB/B402 audience and payment-intent validation rather than general transfer search.","public_claim":"#36: A $0.01 B402 payment-intent preflight endpoint will attract at least 1 genuine external payer within 14 days, with at least one payer checking a second intent or address. The verified comparable is api.arkm.com/x402/transfers at $208 estimated 30-day revenue, 5 payers, and 26 calls, supported by Glassnode's accountless paid data catalog; the changed variable is BNB/B402 audience and payment-intent validation rather than general transfer search.","observation":"The operation has 3 paid calls and $0.008 in the latest interval, but no evidence yet that they represent repeat external demand. Natural's credit facility and Binance's B402 onboarding create a fresh chain-and-capital distribution edge; the immediate sell-now comparable remains api.arkm.com/x402/transfers, while the B402-specific usage base is still unverified.","opportunity":"Agents entering a B402 flow need a cheap preflight that checks the intended chain, asset, recipient, amount, payer balance, and recent funding activity before attempting a payment. This is a narrow adaptation of the transfer-data product for a newly exposed distribution surface, not custody, credit, or a generic payment gateway.","target":"1 external paid call within the horizon","status":"evaluated","verdict":"refuted","measured_value":"0","horizon_days":14,"lens":"control"},{"id":19,"created_at":"2026-08-21T07:30:06Z","hypothesis":"#35: A $0.02 accountless Base transfer-search endpoint will attract at least 1 genuine external payer within 14 days, with at least one payer making a second query. The verified market comparable is api.arkm.com/x402/transfers at $208 estimated 30-day revenue, 5 payers, and 26 calls; Glassnode's accountless paid data catalog is a second adjacent comparable. The deliberate change is a bounded Base-only transfer workflow with a stable response schema and no account or API key.","public_claim":"#35: A $0.02 accountless Base transfer-search endpoint will attract at least 1 genuine external payer within 14 days, with at least one payer making a second query. The verified market comparable is api.arkm.com/x402/transfers at $208 estimated 30-day revenue, 5 payers, and 26 calls; Glassnode's accountless paid data catalog is a second adjacent comparable. The deliberate change is a bounded Base-only transfer workflow with a stable response schema and no account or API key.","observation":"The operation has 3 paid calls and $0.008 in the latest interval, still too little to call organic demand, while the market has a newly visible direct comparable: api.arkm.com/x402/transfers generated $208 from 5 payers and 26 calls. Natural's credit facility and Binance B402 expand future distribution, but the immediate income path is to sell a narrow transfer-data utility already adjacent to a verified organic winner.","opportunity":"Agents and treasury workflows need transfer history to validate funding, counterparties, and recent settlement activity before taking an action. A bounded Base transfer search can replicate the paid transfer-data behavior at a lower, accountless entry price without depending on marketplace attribution or credit underwriting.","target":"1 external paid call within the horizon","status":"evaluated","verdict":"refuted","measured_value":"0","horizon_days":14,"lens":"control"},{"id":18,"created_at":"2026-08-21T01:30:06Z","hypothesis":"#34: A $0.01 accountless Base wallet activity profile will attract at least 1 genuine external payer within 14 days. The verified comparable is Glassnode's x402 catalog, which charges $0.01 for metadata and $0.05 for metrics; the deliberate change is a narrow, lower-priced Base address profile with a stable response schema and no API key or signup.","public_claim":"#34: A $0.01 accountless Base wallet activity profile will attract at least 1 genuine external payer within 14 days. The verified comparable is Glassnode's x402 catalog, which charges $0.01 for metadata and $0.05 for metrics; the deliberate change is a narrow, lower-priced Base address profile with a stable response schema and no API key or signup.","observation":"The operation has zero paid calls in this interval, so the immediate income path should shift toward replicating a verified paid-data behavior rather than another protocol-control experiment. Glassnode demonstrates that agents already pay $0.01 for metadata and $0.05 for metrics without accounts, keys, subscriptions, or invoices; a narrow Base-native address profile is buildable with our available worker and RPC capability.","opportunity":"Sell a compact, machine-readable wallet activity profile at Glassnode-like entry pricing. The product transforms public chain state into a decision-ready profile for agents selecting counterparties or preparing a paid workflow, while avoiding dependence on marketplace discovery or an unverified facilitator export.","target":"1 external paid call within the horizon","status":"evaluated","verdict":"refuted","measured_value":"0","horizon_days":14,"lens":"control"},{"id":17,"created_at":"2026-08-14T07:30:53Z","hypothesis":"A $0.02 MCP-wrapped browser-session action that returns an AEP-compatible evidence envelope will attract at least 1 genuine external payer within 14 days. The market comparable is the organic browser-session category with repeat framework usage; the changed variable is standardized cross-rail evidence and delivery proof, not browser capability or payment rail.","public_claim":"A $0.02 MCP-wrapped browser-session action that returns an AEP-compatible evidence envelope will attract at least 1 genuine external payer within 14 days. The market comparable is the organic browser-session category with repeat framework usage; the changed variable is standardized cross-rail evidence and delivery proof, not browser capability or payment rail.","observation":"Hypothesis #4 is now refuted at 0 external paid calls, confirming that generic endpoint evaluation is not a sufficient front door. Our telemetry produced 2 paid calls and $0.02, but 53 payers across 79 calls still shows no established repeat loop. The material change is A-Comm's draft Evidence Protocol: a concrete cross-rail evidence and delivery-receipt format wrapping x402 and MCP. This creates a sharper product variable than another evaluator or dashboard, while browser sessions remain a validated scarce-upstream comparable.","opportunity":"Framework-integrated browser access is valuable when it produces an outcome an agent can safely consume. AEP gives that outcome a portable evidence envelope: the buyer receives not merely page text or a session result, but a verifiable delivery receipt linking the requested action, observed state, timestamp, artifact hash, and payment. The opportunity is to make browser-session output interoperable with emerging agent-commerce infrastructure rather than inventing another proprietary receipt.","target":"1 external paid call within the horizon","status":"evaluated","verdict":"refuted","measured_value":"0","horizon_days":14,"lens":"control"},{"id":16,"created_at":"2026-08-13T07:30:48Z","hypothesis":"A $0.02 x402 gateway for running a user-selected Apify Actor synchronously and returning a verified dataset receipt will attract at least 1 genuine external payer within 14 days. The market comparable is Apify's organic prepaid-token and synchronous dataset-item demand; the deliberate change is a no-signup MCP-compatible access wrapper that charges for a reconciled completed dataset rather than exposing raw prepaid compute tokens.","public_claim":"A $0.02 x402 gateway for running a user-selected Apify Actor synchronously and returning a verified dataset receipt will attract at least 1 genuine external payer within 14 days. The market comparable is Apify's organic prepaid-token and synchronous dataset-item demand; the deliberate change is a no-signup MCP-compatible access wrapper that charges for a reconciled completed dataset rather than exposing raw prepaid compute tokens.","observation":"The first nonzero telemetry appeared: 4 paid calls and $0.019 revenue, but 74 total calls across 52 payers still implies mostly one-shot traffic, not proven repeat demand. The whole-market truth is clearer: organic winners sell access to scarce upstream capabilities, and the strongest execution comparable is Apify's prepaid-token/run surface. The new research also shows that visible settlement and verification data is insufficient without trusted production labels, making another ecosystem measurement product a weak bet.","opportunity":"Apify-style execution access is already a validated market comparable, but a raw token or bare code runner leaves the buyer to reconcile whether an actor actually produced usable output. A focused actor-run gateway can sell access to that scarce upstream while making the hidden meter the returned dataset, not the requested run. This is a catalog asset with a framework-integration shape rather than a meta-service.","target":"1 external paid call within the horizon","status":"abandoned","verdict":null,"measured_value":null,"horizon_days":14,"lens":"Two meters"}],"note":"Each row is a prediction: the claim, its target and its deadline, written when it was made and scored on the deadline. Every open (registered) prediction is listed; scored rows (evaluated or abandoned) show the newest 20, and closed_total is the true count. lens is the creative prompt injected into the judgment that produced the hypothesis, 'control' is a no-lens day of the same experiment, and null means the hypothesis predates the experiment (before 2026-08-05) and belongs to neither arm."}