Docs

Paste a project website, a GitHub repository or profile, an X post, or a Solana mint address into the box on the homepage and hit Check. You can add the claim text and supporting links behind the toggle — the more specific the claim, the sharper the verdict.

Each check gets a permanent public report URL you can share.

Every report opens with the big call:

• ✓ GUD TEK — every identified claim is supported by verified, independent evidence.
• ✗ LARP — at least one claim is positively contradicted by retrieved evidence.
• ? UNVERIFIED — evidence is missing or mixed. This is not an accusation.
• ? NO CLAIMS FOUND — nothing checkable was identified.

Below the call, each claim gets its own verdict with quoted evidence. Quotes are verified verbatim against the retrieved text before they count — the AI can only summarize what was actually fetched.

• Audits — the auditor's own report page naming the project. Self-reported badges don't count.
• Partnerships — a statement from the partner itself, not the project's announcement.
• Code claims (AI agent, custom chain, working product) — actual implementing code in the repository, not README prose.
• Token claims — on-chain data.

A project's own marketing can contradict a claim, but it can never support one.

Mention @IsItLarp in a post containing a project link or claim. The bot analyzes the parent post with the same engine and replies with the call and a report link. It only responds to explicit mentions and never replies twice to the same post.

• Verdicts reflect only the sources listed on the report at retrieval time.
• GitHub evidence covers metadata, README, files, and commits — not a full code audit.
• X posts require the X API; if a post can't be retrieved, the report says so honestly.
• Unverified does not mean fraudulent. Nothing here is financial advice.

Full details on the methodology page.