agentralabs/agentic-contract

Policies, risk limits, and obligations are evaluated in microseconds through a Rust core with hash-based scope indexing and direct threshold comparison, eliminating the latency of external policy services. The MCP server exposes 38 tools (core + advanced) including predictive capabilities like `violation_precognition` and `risk_prophecy` that forecast breaches before they occur, integrating seamlessly with Claude, GPT, and other LLM-based agents. Python and CLI interfaces provide programmatic and operational access to the same `.acon` contract store, enabling agents to self-check governance compliance and operators to audit policy evolution, budgets, and obligation deadlines in real time.

Available on PyPI.