Make your strongest case before you submit.

Check your proof, make the impact explicit, and catch missing evidence before the triager has to ask.

No account
Code, Solidity and draft-report reviews in your AI chat or coding agent. Use the AI you already have; its usual costs and limits apply.
Free account
All 11 review types on this site, Triager simulation and Prior-art overlap included: 1 a day on your own API key from OpenRouter, OpenAI, Anthropic or another provider.
Spec, a sceptical coral bug in a blue work jacket, holding a magnifying glass.
Meet Spec.A bug with standards.

See why this Critical holds up.

Explore a saved review of the invented Halyard Treasury. The draft claims a 3-of-5 approval bypass; the code and supplied proof support it.

Follow the finding from its claim to the deciding source lines and proof of concept. Then put your own report through the same checks.

Draft report

Example

Title
execute() counts a repeated signature as a new approval, so one owner moves the whole treasury
Draft claims
Critical
Code supports
Critical
Review
execute() counts one owner's repeated signature as threshold approvals; the draft's claims, proof and fix all hold.
Source
input-1/src/HalyardTreasury.sol:76-81
uint256 approvals;
for (uint256 i = 0; i < signatures.length; i++) {
    address signer = _recover(digest, signatures[i]);
    if (isOwner[signer]) approvals++;
}
if (approvals < threshold) revert NotEnoughApprovals(approvals, threshold);
Verdict
Submit2 Oct 2026 · anthropic/claude-opus-5.5
Saved example. Halyard Treasury is an invented protocol.

Run a review

Add your files

Review type

Files

Drop files or a folder here

Source and text files. 120 KB per file. 240 KB, 50 files and 20,000 lines in total.

Paste textCode, a draft report or tool output

Named from what you paste. The review cites it by this name.

Import from GitHubA file, folder, repository or pull request link

Files are read at one exact commit, straight from GitHub to this tab.

Private repository

A read-only token. It goes to api.github.com from this tab and is forgotten on reload.

    Optional. Tell the review what to focus on. Left empty, it uses its standard request.

    Evidence

    Program rules and PoC details

    Check

    When a claim needs correctingTessera example and videos

    A different outcome, backed by the same checks.

    In the invented Tessera Staking example, the reward-theft finding holds, but the evidence does not support a loss of staked principal.

    20 seconds, with narration and captions. The review uses the public, invented Tessera Staking example. Download video
    Read the transcript
    1. 00:00 Critical? Spec has questions.
    2. 00:04 Bounty Operator checks your bug report against the code and proof before you submit.
    3. 00:11 In this example, rewards are affected. Principal is untouched.
    4. 00:16 Try the public example at bounty operator dot com.

    Draft report

    Example

    Title
    Missing checkpoint in stake() lets anyone drain TesseraStaking and lock every staker's funds
    Draft claims
    Critical
    Code supports
    Medium
    Review
    stake() skips settlement, so a fresh staker claims the whole reward reserve; principal is untouched, so the bug is Medium, not Critical
    Source
    input-1/src/TesseraStaking.sol:87-94
    function stake(uint256 amount) external {
        if (amount == 0) revert ZeroAmount();
        _updateGlobal();
        balanceOf[msg.sender] += amount;
    
    Verdict
    Rewrite, then submit2 Oct 2026 · anthropic/claude-opus-5.5
    Saved example. Tessera Staking is an invented protocol.

    Follow the Tessera review in the workbench.

    The 54-second walkthrough follows this saved example from the draft to the cited code and downloadable packet.

    54 seconds, no sound. Actual screens from the saved public example. Tessera Staking is an invented protocol. Download video
    Read the transcript
    1. 00:00 See what needs fixing before you submit. The tour opens the public Bounty Operator example.
    2. 00:04 Add your evidence: the draft report, source code, proof and known issues. The example contains four files.
    3. 00:12 Choose your review route: your own API key or a chat subscription. This tour loads the saved example rather than running another model review.
    4. 00:18 Read the verdict: rewrite, then submit. The reward-theft finding is supported, but the Critical impact claim is not.
    5. 00:27 Inspect the claim table. The review marks overstated claims and links its objections to the supplied file and line.
    6. 00:35 Open the cited source. Line 115 of TesseraStaking.sol rejects a payout above rewardReserve, limiting this payout to the reward reserve.
    7. 00:43 Download the review packet to keep the verdict, source citations and SHA-256 file manifest together.
    8. 00:50 Try the public example at bountyoperator.com and follow the evidence yourself.

    Working in a coding agent? Watch the MCP walkthrough.

    Every review ends in one of five verdicts.

    One decision, with the file and line behind it. The draft stays yours to rewrite.

    Paydirt: 22 models on 18 held pairs

    Benchmark release 2026-10

    1. GPT-6 Astra94.4 Paydirt Score$0.52 a run
    2. GPT-6.1 Sol94.4 Paydirt Score$0.108 a run
    3. Kimi K388.9 Paydirt Score$0.22 a run
    4. GPT-5.6 Terra88.9 Paydirt Score$0.433 a run
    See the leaderboard

    Free for 1 review a day. US$10 a month for unlimited.

    Operator includes US$2 in Bounty Operator API credit per paid month. Other model providers bill usage separately.

    Free

    US$0

    No card or email.

    • 1 review a day, any of the 11 review types, on your API key with a free account
    • Triager simulation, Prior-art overlap and 6 more checks a chat app does not run
    • Unlimited copy-paste reviews in ChatGPT or Claude (3 review types), no account
    • Unlimited reviews by your coding agent’s own model (3 review types), no account

    Operator

    US$10per month

    Everything in Free, plus:

    • Unlimited reviews, 4 at once
    • US$2 in Bounty Operator API credit per paid month
    • Gauntlet: eight stages on one finding, then a verdict
    • Panel review: compare 2 to 4 models

    Billed monthly. Cancel any time in the Stripe portal and keep access until the paid period ends. Full plan comparison · Terms

    Questions

    I already pay for my model. Why pay for this?
    The US$10 monthly plan pays for unlimited reviews, the Gauntlet, Panel review and 4 reviews running at once. It includes US$2 in Bounty Operator API credit per paid month. Other model providers bill usage to your own key. Pricing
    Can I use my ChatGPT or Claude subscription?
    Yes, for 3 review types: code security, Solidity and draft reports. Copy the prompt into your chat and paste the answer back here. The other 8 review types need an API key and a free account. Model options
    Where do my files go?
    Your files and API key pass through our server to your provider. The server adds the review instructions and stores no files, prompts, keys or results. A copy-paste prompt stays in your browser until you copy it. Privacy details
    Which models can I use?

    Choose a model from OpenRouter, Anthropic, OpenAI, Google Gemini, xAI, DeepSeek, Mistral, Groq, Novita AI, Nebius Token Factory, Infron, OrcaRouter or Bounty Operator, on your own API key.

    How do I cancel or get help?
    Cancel in the Stripe billing portal. Access continues until the end of your paid period. Contact support@bountyoperator.com for access problems or billing help. Refund terms

    Import from GitHub

    The prompt your model receives

    
    

    Cited lines

    Local history

    Off by default. When on, each finished review is saved in this browser as its packet and manifest. Source files and keys are never saved. Nothing is uploaded.

    Sign in

    Account

    Confirm it's you