GustyCube/anchor
Open protocol for deterministic, offline verification of AI agent authorization.
Anchor provides a standard, secure way to confirm that an AI agent is truly authorized to perform an action, such as making an API call or updating a database. It works by having the system that grants permissions create a signed "Capability" that the agent attaches to every action. The service receiving the action can then verify this permission locally and instantly, without needing to contact the original issuer, ensuring fast, auditable, and reliable decision-making. Operations engineers, security architects, and platform developers managing AI agent interactions with critical services will find this useful.
Use this if you need a decentralized, auditable, and highly reliable method to verify AI agent authorization for actions like API calls or data writes, without introducing network latency or dependencies on a centralized authorization service.
Not ideal if your system already has a robust, low-latency, and auditable centralized authorization solution that meets all your security and operational requirements for AI agents.
Stars
25
Forks
—
Language
Go
License
—
Category
Last pushed
Mar 22, 2026
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/agents/GustyCube/anchor"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
heshengtao/super-agent-party
⭐ All-in-one AI companion! Super Agent Party = Self hosted neuro sama + openclaw! ⭐...
darfaz/clawmoat
🦀 Security moat for AI agents. Runtime protection against prompt injection, tool misuse, and...
dataelement/Clawith
OpenClaw for Teams
romanklis/openclaw-contained
TaskForge runs AI agents in sandboxed Docker containers with capability-based security. Agents...
scottgl9/LeanClaw
LeanClaw is a high-efficiency, security-first AI assistant runtime built for fast local...