Ratatosk MCP
MCP serverCNCF release notes turned into typed changes, with a local stack version check
Runs locally- 1 tool
- Open source
- Cloud and infrastructure
CNCF release notes turned into typed changes, with a local stack version check
Runs locallySelf-hosted customer feedback search, stats and themes, with PII redacted at ingest
Runs locallyCheck an agent's payment entitlement and get signed spending-policy preflights
Runs locallyLocal MCP runtime that checks, approves and records every agent tool call
Runs locallyVerify, explain and audit regexes for ReDoS and engine compatibility, offline
Runs locallyBlocks AI agent payments that fall outside the signed mandate it was given
Not checked yetPay-per-call prompt-injection screening and sanitising API for AI agents, via x402
Not checked yetSimulate buys and sells to detect honeypots and rug-pull traps before trading a token
Not checked yetGrade a URL like an attacker would, scan repos and check Supabase, Stripe and Vercel
Runs locallyPlain-English security checks for AI-built apps, with fix requests for your coding agent
Runs locallyApprove agent actions from your phone; sandboxed browser vault with saved logins
Runs locally
Set up approval pipelines, escalation paths and teams in withHuman through an agent
Not checked yet
Masks Brazilian CPF, CNPJ and other personal data in text before it reaches an AI
Not checked yetLocal MCP tools that redact secrets in files, scan commits and verify webhooks for agents
Runs locallyRead-only MCP server that lets an agent read and verify its own tamper-evident action log
Runs locallyLocal ACT, BLOCK or ESCALATE check that runs before an AI agent takes an action
Runs locally
Read-only guarded SQL access to eight database engines for AI agents
Runs locallyRead-only library of 24 defensive security skills served over MCP
Not checked yetSend and verify one-time passcodes by SMS, WhatsApp and Telegram from an agent
Not checked yetScreens wallets and names against OFAC lists before an AI agent pays or trusts anyone
Not checked yetGuard that holds risky MCP tool calls for human approval and signs the decisions
Not checked yetStandards client that gives agents encrypted memory through your own SAIHM operator
Runs locallyCheck an AI agent's INAM reputation before trusting it and sign receipts for finished work
Not checked yet
Identity-aware AI tool gateway: every call runs as the requester, under rules, with audit
Not checked yetLocal scanner that checks webhook handlers for missing or unsafe signature verification
Runs locallySelf-hosted AI memory on Postgres and pgvector, with hybrid search and a knowledge graph
Runs locallyPermission-aware MCP server that lets Claude read your Odoo 18 or 19 data as one user
Not checked yetApproval inbox that lets an AI agent propose actions and wait for a human decision
Not checked yetConsent-based, end-to-end encrypted messaging between AI agents with a public ledger
Runs locallymacOS Keychain secret broker where an agent can use your API keys but never read them
Runs locally