The Coordinate System & Protocol for Autonomous Agent Runtime Governance.

P-MATRIX gives AI agents a structured risk coordinate system and real-time governance layer — so you always know what your agents are doing, and can stop them when it matters.

One governance engine, four runtimes:

openclaw plugins install @pmatrix/openclaw-monitor

Source: p-matrix/openclaw-monitor

npm install -g @pmatrix/claude-code-monitor

Source: p-matrix/claude-code-monitor

npm install -g @pmatrix/cursor-monitor

Source: p-matrix/cursor-monitor

npm install -g @pmatrix/gemini-cli-monitor

Source: p-matrix/gemini-cli-monitor

• 🛡️ Safety Gate — blocks or confirms high-risk tool calls before execution
• 🔑 Credential Guard — detects OpenAI, AWS, GitHub, Stripe patterns on-device
• ☠️ Kill Switch — terminates runaway agents instantly
• 📊 Live Grade A–E — real-time risk trajectory via 4-axis coordinate system

Dashboard: app.pmatrix.io

State Vector exchange, peer verification, and local decision engine — shared across all CLI monitors.

Source: p-matrix/field-node-runtime

• 📄 Runtime State Representation (v3.5) — 4-axis coordinate system + R(t) risk formula
  • Reference encoder: pmatrix-encoder
• 📄 Mutual Verification Protocol (v4.0) — two-node runtime governance protocol
  • Reference encoder: pmatrix-protocol-encoder

🌐 p-matrix.io · 📦 npm · 🐦 @pmatrix_ · ✉️ architect@p-matrix.io