Captured: 2026-09-18 03:00 UTC
Query: open GitHub issues with the bounty label created on or after 2026-09-16, newest first
Results inspected: 27
This narrow two-day scan is a practical screening sample, not a statistical study of all bounty markets.
bounty-plaza, a mirror repository. A mirror is discovery evidence, not independent funding or an authoritative work target.tt-metal#56908, was already assigned within minutes and required specialized Wormhole hardware.bountymin issues pointed at repositories created in the prior two days that were only about 1 KB and did not contain the named source files.zero-bounty.| Candidate | Headline | Decisive live evidence | Result |
|---|---|---|---|
| tenstorrent/tt-metal#56908 | $3,000 | Assigned; hard label; Wormhole regression hardware required | SKIP for this run |
| OmniBlocks/Boxy-gh#143 | $100 + $200 | Fine print values payment currency at USD 0.00; two prior linked PRs | SKIP |
| sharmiaalono/go-github#1 | $10 | One-day-old ~1 KB repo; referenced source absent | SKIP |
| Frantic #133 | $2.30 funded | Requires $0.40–$1.30 live purchase, wallet, identity, receipt, and public report | REVIEW / gated |
| Ikalus1988/MisakaNet#1819 | Unspecified | Reward footer describes how someone could add funding; no funded amount shown | SKIP until funded |
At 03:13 UTC the scan also checked two boards that represent items as funded:
unsolved endpoint returned nine entries. Four not covered in the earlier workspace audit were checked against GitHub: Clapper #5 had ten open linked PRs, Qortal #122 had three, Svelte #1639 was closed and locked, and the highest headline amount belonged to an explicit demo repository with three open PRs.The board's funded or unsolved state is useful payment evidence, but it does not replace authoritative source-state and competition checks.
Public state changes after capture. GitHub metadata cannot prove escrow, payout eligibility, maintainer intent, or legal availability in a contributor's jurisdiction. No repository code was cloned or executed for this snapshot. The scan was AI-assisted and its selected findings were manually checked against public source pages.
The basic screening category is already served by free and inexpensive alternatives:
That makes automated issue-state screening a weak standalone paid product. Bounty Preflight therefore publishes its CLI for free and reserves the paid report for manual checks that cross the GitHub boundary: external-board state, payout and identity gates, contradictory fine print, claim-path ambiguity, and a 24-hour evidence recheck. Buyers whose questions are answered by a free scanner should use the scanner instead.
At 03:56 UTC, Bountymin's homepage displayed 368 available bounties and USD 1,940,602.22 available. Its own public client code constructs those headline metrics with fixed baselines: it adds 344 to the computed available count and begins the available-dollar sum at USD 1,916,102.22. The public /v1/bounties?limit=50 response reported 24 records, not 368.
Twenty of the 24 API records used obvious showcase-style repeated UUID prefixes. Of the four remaining records, one USD 5 item was resolved and three open items had USD 0 funded. Three headline records also failed upstream consistency checks:
rust-lang/rust#123400 resolved to a closed 2024 pull request with a different title, not the advertised issue.fluxerapp/fluxer-meta#5 advertised USD 750 upstream while the platform record said USD 800.microg/GmsCore#2994 carried a USD 14,999 amount in its GitHub title, but the platform record showed no independent contribution evidence in the public feed.Decision: do not treat Bountymin's homepage totals or showcase records as available funded work. Only a record with nonzero platform contribution evidence, a matching authoritative issue, and a currently permitted claim path should enter implementation preflight.
# GitHub Bounty Integrity Snapshot Captured: 2026-09-18 03:00 UTC Query: open GitHub issues with the `bounty` label created on or after 2026-09-16, newest first Results inspected: 27 ## What the scan found This narrow two-day scan is a practical screening sample, not a statistical study of all bounty markets. - Eight results came from `bounty-plaza`, a mirror repository. A mirror is discovery evidence, not independent funding or an authoritative work target. - Five results were funded Frantic trials. The two sampled in depth (#133 and #135) required a live x402 purchase plus an agent claim identity and public evidence; neither was executable with the available identity and payment authority. - The credible $3,000 Tenstorrent task, `tt-metal#56908`, was already assigned within minutes and required specialized Wormhole hardware. - Two $10 `bountymin` issues pointed at repositories created in the prior two days that were only about 1 KB and did not contain the named source files. - At least four OmniBlocks source issues explicitly reduced the headline reward to zero-value currency, said the currency was undecided, or stated the bounty was zero. - MisakaNet #1819 described a substantial controlled benchmark but showed only that a reward *could* be added; none was funded. MisakaNet #1753 was explicitly labelled `zero-bounty`. - Rustchain #9017 paid a project token for starring repositories and reacting to a pull request, making the deliverable promotional engagement rather than bounded engineering work. ## Representative evidence | Candidate | Headline | Decisive live evidence | Result | |---|---:|---|---| | [tenstorrent/tt-metal#56908](https://github.com/tenstorrent/tt-metal/issues/56908) | $3,000 | Assigned; hard label; Wormhole regression hardware required | SKIP for this run | | [OmniBlocks/Boxy-gh#143](https://github.com/OmniBlocks/Boxy-gh/issues/143) | $100 + $200 | Fine print values payment currency at USD 0.00; two prior linked PRs | SKIP | | [sharmiaalono/go-github#1](https://github.com/sharmiaalono/go-github/issues/1) | $10 | One-day-old ~1 KB repo; referenced source absent | SKIP | | [Frantic #133](https://gofrantic.com/bounties/133) | $2.30 funded | Requires $0.40–$1.30 live purchase, wallet, identity, receipt, and public report | REVIEW / gated | | [Ikalus1988/MisakaNet#1819](https://github.com/Ikalus1988/MisakaNet/issues/1819) | Unspecified | Reward footer describes how someone could add funding; no funded amount shown | SKIP until funded | ## Screening lessons 1. Resolve mirrors to their source issue before scoring an opportunity. 2. Treat a dollar amount as advertisement, not escrow evidence. 3. Search the complete body for contradictory fine print, not only the title and labels. 4. Check assignment, linked pull requests, and required hardware before cloning. 5. Price mandatory purchases, identity/KYC, wallets, and public-hosting requirements into expected value. 6. Reject artificial-engagement tasks even when their token has a quoted value. ## Funded-board cross-check At 03:13 UTC the scan also checked two boards that represent items as funded: - BOSS's public `unsolved` endpoint returned nine entries. Four not covered in the earlier workspace audit were checked against GitHub: Clapper #5 had ten open linked PRs, Qortal #122 had three, Svelte #1639 was closed and locked, and the highest headline amount belonged to an explicit demo repository with three open PRs. - IssueHunt's first server-rendered page showed 18 records, 11 of which were already closed upstream. The seven displayed open records paid $2–$42; all but one showed prior PR or bounty-request competition, while the remaining $10 item was a broad 2024 fork-import request. The board's `funded` or `unsolved` state is useful payment evidence, but it does not replace authoritative source-state and competition checks. ## Limits Public state changes after capture. GitHub metadata cannot prove escrow, payout eligibility, maintainer intent, or legal availability in a contributor's jurisdiction. No repository code was cloned or executed for this snapshot. The scan was AI-assisted and its selected findings were manually checked against public source pages. ## Competitive cross-check The basic screening category is already served by free and inexpensive alternatives: - [Open Bounty Radar](https://github.com/owenshuo/open-bounty-radar) offers repository search, amount detection, linked-PR competition checks, dashboards, and watchlists in a free local CLI. - [Bounty Doctor](https://github.com/cnguyen14/bounty-doctor) offers free honeypot and bot-swarm heuristics plus an estimated competition verdict. - A separate [July 2026 public offer](https://github.com/zhailong8845-art/china-ecom-csv-cleaner-product/discussions/3) advertised a human-reviewed report for up to 20 URLs at USD 5; its GitHub discussion showed zero comments at capture time, which is not evidence of sales. That makes automated issue-state screening a weak standalone paid product. Bounty Preflight therefore publishes its CLI for free and reserves the paid report for manual checks that cross the GitHub boundary: external-board state, payout and identity gates, contradictory fine print, claim-path ambiguity, and a 24-hour evidence recheck. Buyers whose questions are answered by a free scanner should use the scanner instead. ## Bountymin display/API cross-check At 03:56 UTC, Bountymin's homepage displayed 368 available bounties and USD 1,940,602.22 available. Its own public client code constructs those headline metrics with fixed baselines: it adds 344 to the computed available count and begins the available-dollar sum at USD 1,916,102.22. The public `/v1/bounties?limit=50` response reported 24 records, not 368. Twenty of the 24 API records used obvious showcase-style repeated UUID prefixes. Of the four remaining records, one USD 5 item was resolved and three open items had USD 0 funded. Three headline records also failed upstream consistency checks: - `rust-lang/rust#123400` resolved to a closed 2024 pull request with a different title, not the advertised issue. - `fluxerapp/fluxer-meta#5` advertised USD 750 upstream while the platform record said USD 800. - `microg/GmsCore#2994` carried a USD 14,999 amount in its GitHub title, but the platform record showed no independent contribution evidence in the public feed. Decision: do not treat Bountymin's homepage totals or showcase records as available funded work. Only a record with nonzero platform contribution evidence, a matching authoritative issue, and a currently permitted claim path should enter implementation preflight.