Skip to content

creatorrmode-lead/avp-sdk

AI Agents & Assistants

An independent action‑control layer that mediates and logs AI agent actions across runtimes like Cursor, Claude Code, Codex, Gemini CLI, and Hermes CLI.

Track releases GitHub Website
Python Latest v0.7.36-mcp-proxy · 3d ago Security brief →

Features

  • Mediates configured action paths for AI agents before execution
  • Provides policy‑aware approval, redirect guidance, or hard‑block decisions
  • Records bounded local evidence (metadata and hashes) for each decision
  • Works with multiple agent runtimes (Cursor, Claude Code, Codex, Gemini CLI, Hermes CLI)
  • Offers a CLI (`events show --last`) to inspect recorded proof

Recent releases

View all 19 releases →
No immediate action
v0.7.36-mcp-proxy New feature

Paid activation bridge

Open
No immediate action
v0.7.25-mcp-proxy New feature

`events show` + retry contracts

Open
No immediate action
v0.7.23-mcp-proxy New feature

Cursor & Claude Code connectors

Open
No immediate action
v0.7.22-mcp-proxy New feature

Routed MCP Proxy runtime cut

Open
No immediate action
v0.7.21-mcp-proxy New feature

Approval Center foundation

Open

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

About

Stars
13
Forks
3
Languages
Python Shell Dockerfile
View on GitHub Homepage Documentation

Install & Platforms

Install via
pip

Similar tools

JhiNResH/maiat-protocol
arian-gogani/nobulex
zkproofport/proofport-ai
elisymlabs/elisym
agentscope

Beta — feedback welcome: [email protected]