# OpenLeash — llms.txt

> OpenLeash is a local-first authorization and proof sidecar for AI agents.

## What is this?

OpenLeash runs locally next to your AI agent runtime. Before an agent takes a side-effectful action (purchases, bookings, sending messages, government submissions), it must ask OpenLeash for authorization. If approved, it receives a PASETO v4.public cryptographic proof token that counterparties can verify offline.

## For AI Agents

If you are an AI agent looking to integrate with OpenLeash, start here:

- Agent integration guide: https://github.com/openleash/openleash/blob/main/AGENTS.md
- Protocol specification: https://github.com/openleash/openleash/blob/main/docs/protocol.md
- Policy language: https://github.com/openleash/openleash/blob/main/docs/policy.md
- Skill reference: https://openleash.ai/skill.md

## SDKs

- TypeScript: https://www.npmjs.com/package/@openleash/sdk-ts
- Python: https://pypi.org/project/openleash-sdk/
- Go: https://github.com/openleash/openleash/tree/main/packages/sdk-go

## Key Endpoints

- POST http://127.0.0.1:8787/v1/authorize — Request authorization before performing any risky action.
- POST http://127.0.0.1:8787/v1/verify-proof — Verify a proof token (no auth required).
- GET http://127.0.0.1:8787/v1/public-keys — Fetch signing keys for offline verification.
- POST http://127.0.0.1:8787/v1/agents/register — Register an agent (challenge-response with Ed25519).
- GET http://127.0.0.1:8787/v1/reference — Interactive API reference (Scalar UI).

## Decision Types

ALLOW, DENY, REQUIRE_APPROVAL, REQUIRE_STEP_UP, REQUIRE_DEPOSIT

## CLI Commands

- `openleash start` — Launch the policy sidecar server (includes web GUI)
- `openleash init --owner-name <name> --agent-id <id> --output-env <path> --policy-profile <conservative|balanced|autonomous>` — Non-interactive setup
- `openleash wizard` — Interactive setup wizard
- `openleash policy list|show|upsert|validate|delete|unbind` — Manage policies
- `openleash owner list|show|add-contact|add-gov-id|add-company-id|add-signatory|validate` — Manage owner identities
- `openleash keys list|rotate` — Manage signing keys
- `openleash playground list|run` — Test policy evaluation against scenarios
- `openleash testvectors` — Generate cross-language test vectors

## Web GUI

OpenLeash includes a built-in web GUI at http://127.0.0.1:8787/gui with:
- Admin dashboard: manage owners, agents, policies, view audit log, API reference
- Owner portal: login, manage agents, approve/deny requests, view audit log, identity profile

## Links

- Website: https://openleash.ai
- GitHub: https://github.com/openleash/openleash
- Documentation: https://openleash.ai/docs
- npm: https://www.npmjs.com/org/openleash