Vacant
Accountability layer for AI agents: run the client's executable acceptance tests, gate delivery, sign every step into a verifiable receipt. Pre-registered, replicated 5x, cross-benchmark.
Warning. Serious findings were identified. Review the full report before connecting. Grades reflect the full trust pyramid: code, verification depth, and reputation. New projects cap at C until adoption is earned.
How we verified
Verified is a snapshot. Live keeps it current, and builds your track record.
⚡ Connect GitHub → continuous verification on every pushwhy connect →Who stands behind it
cosmopig
Source: github_repo_search
Claim it to get a verified publisher badge, a free copy of our full audit findings, and direct contact for any high-priority issues we find. Or connect your repo for our deepest verification, Live Monitored: read-only, revoke anytime. What we access →
Install from
The grade above is for the source repository. Registries can serve a different version, so we mark the ones we were not able to read.
These names and descriptions are the publisher's own, read from the source code. We print them as written. Our assessment is the findings above, not this list.
delegateTHE PREFERRED PATH for any coding subtask with an objective check. Instead of writing the code yourself, hand it to Vacant's trusted, accountable resident ecosystem: the task is routed (by reputation) to a resident, generated on Vacant's local model with relevant memory injected, cross-reviewed by K…
trust_cardFetch the FULL trust card JSON for a prior delegate call (by its task_id): deliverer identity + credit/observations/flags, every signed peer review, audit status, chain head, and host signature. Returns a JSON error string if unknown.
receiptFetch the full signed delegation receipt and its exact immutable trust card by request_id. The receipt binds the request, objective check, risk, delivered answer, trust card, resident identity, stream and chain head. Returns JSON error text when unknown.
residentsShow the resident roster: each resident's tier, credit score, observation count, deliveries, episodes, chain integrity, and flags. Flags render honestly — INSUFFICIENT_DATA (n<30 observations) and PROBATION are shown as-is, never hidden.
reportHuman arbitration feed-back (the strongest label): report the real outcome of a delegated task by task_id. verdict e.g. "PASS"/"FAIL"/"FAULT"; a fault verdict emits a slash event. Returns a JSON ack.
scoreboardPaired trust OFF vs ON evidence: pass counts, call counts (cost), and the paired pass-rate delta. Every delegate call — trust on or off — is one trial recorded here.
verify_fixStrengthen YOUR OWN answer to a task that has an OBJECTIVE check. vacant runs a verify-fix loop on its local model: generate -> run `check` -> if it fails, retry with feedback (up to k times) -> return the first answer that PASSES, plus a signed attestation you can show as proof. Use this ONLY when …
VACANT_MCP_MODELVACANT_MCP_BASEVACANT_MCP_APIVACANT_MCP_API_KEYVACANT_API_KEYVACANT_MCP_ROOTVACANT_TRACEVACANT_LOCAL_APIVACANT_LOCAL_MODELVACANT_EVALPLUS_PATHVACANT_LCB_PATHVACANT_ENDPOINTCLINE_KEYSEQ5_ANALYZE_MUTANTMUTANTR460_MUTANTR529_MUTANTVACANT_GAIN_APICLINE_KEY_INDICESR477_MUTANTR450_MUTANTR455_MUTANTR466_MUTANTR479_MUTANTR480_MUTANTR482_MUTANTR483_MUTANTR484_MUTANTR485_MUTANTR486_MUTANTR487_MUTANTR487B_MUTANTTool annotations
No tools have read-only/destructive annotations
Add readOnlyHint or destructiveHint annotations to every tool so hosts can warn users before invoking.
All four hints declared on every tool
7/7 tools missing one or more hints — delegate (missing: readOnlyHint, destructiveHint, idempotentHint, openWorldHint); trust_card (missing: readOnlyHint, destructiveHint, idempotentHint, openWorldHint); receipt (missing: readOnlyHint, destructiveHint, idempotentHint, openWorldHint), +4 more. OpenAI's directory rejects tools where any of the four hints are missing or non-boolean.
For every tool, set all four hints (readOnlyHint, destructiveHint, idempotentHint, openWorldHint) to explicit true/false values that match the handler’s actual behaviour.
No hardcoded API keys
2 live-looking API keys in source: 2 AWS access key
Move secrets to environment variables (process.env.X) or your secret manager.
Claim the listing to review these findings one by one and send us a correction where you disagree, straight to the team. Claiming also means we tell you when the grade moves, and reach you first if we find anything urgent.
[](https://m8ven.ai/mcp/cosmopig/vacant)?variant=verified from the URL.Vetting this one by hand? Tool Check is an MCP that scores other MCPs. Add it once and ask Claude, ChatGPT, or any MCP client to grade a server, surface CVEs, check the publisher, and suggest safer alternatives — before you install.
https://m8ven.ai/api/mcp/tool-check