A local AI Agent tool governance gateway: it does not prevent prompt injection, but prevents the high-risk tool-call consequences after injection succeeds (policy / approval / audit / bidirectional MCP governance).
Latest release v0.4.2 — AgentToolGate v0.4.2 · 17 Aug 2026
2 files for Codex, OpenCode and Claude Code: AgentToolGate AGENTS.md, AgentToolGate CLAUDE.md — 807 tokens loaded in every session.
AGENTS.md A 802 tok CLAUDE.md A 5 tok These files are aki0225/AgentToolGate's own configuration — they tell Codex, OpenCode and Claude Code how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.