Open Protocol for Causal Governance in Agentic AI

PIC closes the causal gap : when untrusted inputs (prompt injection, user text, web pages) influence high‑impact side effects (payments, exports, infra changes), PIC forces a machine‑verifiable contract between what the agent claims and what evidence actually backs it .

Guardrails constrain what the model says . PIC constrains what the agent is allowed to do (side effects) based on verifiable provenance + evidence .

Deterministic evidence verification : evidence IDs can point to a real artifact and be validated via SHA‑256 .

Signature verification (Ed25519) so approvals can be endorsed by trusted signers (CFO, internal service, billing system) without shipping the raw artifact .

Signature evidence is verified against a local keyring file.

Lightweight it comes with LangGraph, MCP and OpenClaw integrations.