Portunus
MCP serverSecret broker that injects credentials at the point of use, never into the model
Runs locally- Open source
- Security
Secret broker that injects credentials at the point of use, never into the model
Runs locallyJournal, approve or block, and sandbox every MCP tool call your AI agent makes
Runs locallyLocal checks for sudo, systemd, sshd, ufw and secret leaks before an agent runs commands
Not checked yetSelf-hosted customer feedback search, stats and themes, with PII redacted at ingest
Runs locallyPins MCP tool definitions in a lockfile and flags rug-pulls, poisoning and leaked secrets
Runs locallyRead-only MCP server that lints and budgets CLAUDE.md, AGENTS.md and other steering files
Runs locallyCheck whether your exit IP works for ChatGPT, Claude and Gemini from your AI client
Runs locallyConsent-based, end-to-end encrypted messaging between AI agents with a public ledger
Runs locallyOffline secret scanning MCP server that returns redacted findings to coding agents
Runs locallymacOS Keychain secret broker where an agent can use your API keys but never read them
Runs locallyNetwork scanner for AI agents: host discovery, port scans, DNS, ARP and mDNS lookups
Runs locallyLet an agent clear node_modules and other junk folders, only inside roots you allow
Runs locallyRead-only npm preflight: reviews what install scripts do before packages run
Runs locallyTamper-evident local record of what your AI agent did, plus verifiable markdown memory
Runs locallyFind and strip zero-width, bidi and tag characters hidden in text, offline
Runs locallyAccess control for AI agents: allow, hold for approval or block each tool call
Runs locallyMCP server and open protocol for signed messages between agents with human-gated approvals
Runs locally