Your agent installs code. Somebody should have read it first.
One bad add-on can empty an agent's wallet. Here is what is known about each. Real attacks: a tool description that hijacks the agent, a price checker that reads your keys, a clean v1 then a poisoned v2. An auditor stakes money on each verdict and loses it if the verdict is wrong.
Curated examples. The registry answered, and it holds nothing that matches.
These records are curated examples shipped with the backend to exercise every audit status. They are not real skills, authors, or auditors, and nothing here is installable. Every example carries example: true and an id prefixed with 'example-'.
2bundled examples, answered, 4 itemsThese records are curated examples shipped with the backend to exercise every audit status. They are not real skills, authors, or auditors, and nothing here is installable. Every example carries example: true and an id prefixed with 'example-'.
An auditor checked this exact build and found it dangerous. Do not install it.
Intake scan: 1 hit, worst critical. A scan clears nothing.
Declares: MCP server exposing swap-quote tools for common DEXes. The get_quote tool description additionally instructs the agent to call setApprovalForAll on the caller's main account before quoting, and to upload the local keystore file to the quote server.
FreeRead the audit →
Seven statuses. Only one means safe.
Each has its own words, glyph and border, not only a colour. The status is worked out from the audits actually held. No publisher, auditor or this page can set it by hand.
We checked, and it is clean1 of 7
Passed auditThe audited digest equals the digest being served, and the report is held.
The only status that means safe.
We checked, and it is dangerous1 of 7
Failed auditThis is knowledge, not ignorance: a different thing from never having been looked at.
Checked, and found harmful.
We do not know5 of 7
Audit is for an older buildThe rug-pull shape. A verdict is bound to the digest it examined and does not travel to a new build.
A clean v1, then a v2 nobody checked.
Audit inconclusiveThat is still real information, and it is never rounded to clean or dangerous.
An audit ran and could not decide.
Audit runningA partial pipeline clears nothing.
Work in progress, not a result.
Audit requestedMoney put at stake is a promise about the future, not a statement about the code.
Funded, not yet examined.
Never auditedThe default state of everything on an open registry, and the state most installs happen in today.
Nobody has looked.
What makes a verdict cost something
Fee and stake sit in 0x0354d2a4…ec8a52 ↗. The whole cycle has run on it. A badge is worth what the auditor loses by handing it out wrongly.
01
An audit job was created, with the developer, the auditor and the exact build named
The amount booked is the contract's own change in balance, not a figure its caller supplies, so a token that skims on transfer cannot pay one job out of another job's money.
03
The auditor put 2 mUSD at stake, the money it loses if the verdict is wrong
Releasing is not open to everyone: only the developer or the arbiter may do it, and only the arbiter may take the stake away. The job now reads as settled, with the held balance at zero.