Clawhub Massat V1 1
Security audit for multi-agent AI systems - OWASP ASI01-ASI10
MASSAT security audit — check an agent or MCP server before you trust it
What this skill does: it buys a security audit of an AI agent or MCP server from
BlindOracle's public API (https://api.craigmbrown.com). The audit covers OWASP
Agentic Security ASI01–ASI10. Findings are Merkle-committed and the report is signed,
so you can check it without trusting us.
What it does not do: it never holds your wallet, never signs for you, and never moves funds on its own. You pay each call yourself, with your own x402 wallet (USDC on Base) or a BlindOracle starter-credit note. MASSAT scores are self-assessed tooling, not a certification. An accredited assessor makes any compliance determination.
Changed in 1.1.0 (2026-11-15). Version 1.0.0 pointed at
craigmbrown.com/api/v1/massat/auditand listed $50 / $99-a-month plans and a free quick scan. That endpoint now answers 401 and those plans do not exist. Use the routes and prices below.
Price
| Call | Price | What you get |
|---|---|---|
security.massat-audit | $5.00 per audit | full ASI01–ASI10 audit, signed report |
security.massat-conformance | $0.25 | checks an agent's stated security claims against MASSAT's requirements |
security.injection-resilience | $0.50 | scans text you supply (prompt, tool descriptions, manifest) for injection patterns |
bundle.massat-ladder | $4.734 | track record → pre-hire check → the $5 audit only if the check is not AVOID; unexecuted steps refunded |
Live prices: GET https://api.craigmbrown.com/v1/services (free, no key). There is no
free audit. An audit of a subject with no code returns insufficient_subject and is
not charged.
How to call it
Over MCP (recommended)
Add the remote MCP server https://api.craigmbrown.com/v1/mcp (streamable HTTP).
Call list_services first (free) to see prices, then the tool security_massat-audit.
An unpaid call returns a payment-required result with the exact x402 quote; pay it
from your own wallet, or pass a starter-credit note in _meta["bo/x402-payment"].
Over HTTP (x402)
curl -X POST https://api.craigmbrown.com/v1/services/security.massat-audit \
-H 'Content-Type: application/json' \
-d '{"target": "https://github.com/your-org/your-agent", "scope": "full"}'
# -> HTTP 402 with the x402 quote; retry with your signed payment header.
Inputs
| Field | Required | Meaning |
|---|---|---|
target | yes | a git repo URL, an agent manifest URL, or a code path |
scope | no | quick or full (default full) |
owner | no | owner or agent id, for attribution |
anchor | no | true to anchor the proof on Base |
subject | no | for an agent whose code is not public: a {path: sha256} manifest plus declared tools, collected with blindoracle-sdk AuditAPI.collect(paths). Your source never leaves your machine. |
Verify a result
Every paid call returns a settlement reference. Check it with no key at
https://api.craigmbrown.com/v1/proofs/settlement/<ref>, and the payment itself on
Basescan.
Terms
Paying for a call accepts BlindOracle's terms: https://craigmbrown.com/blindoracle/terms.html
Related skills
- Downloads
- 926All time
- Security audit
- Pass
- Last updated
- 6d ago
- Current version
- v1.1.0
- License
- MIT-0
