Back to Gap Map Product / UX / Agent identity & authorization

Permit0

Permit0
open source / Overall score: 1.8

Permit0 is a policy engine that decides AI agent actions before they run. It maps each tool call onto a published taxonomy of agent actions, scores it against risk rules, and allows it, blocks it or sends it to a person, writing each verdict to a signed audit chain. It hooks into Claude Code and OpenClaw, and ships an email pack for Gmail and Outlook.

Openness

5 medium confidence
5.0
license
Apache-2.0 (engine, command-line tool, dashboard, taxonomy and email pack)
source
public
core features withheld
no — the 342-path public tree at c5e8f7d has one Apache-2.0 LICENSE and no ee/ or enterprise/ path, and permit0.com, which names no paid edition, offers the full SDK free with no rate limits; docs/enterprise-strategy.md only plans a private enterprise fork, which no gate in the published code anticipates
commercial
planned permit0-enterprise edition (a private fork with commercially licensed crates, described in the public repository)

Permit0's engine, command-line tool, dashboard and email pack are published under Apache-2.0 and run offline on the user's own machine, and nothing in the published code is held back for a paid tier. The project's planning document sketches a private enterprise edition with SAML sign-on, fine-grained roles and multi-tenancy, but no such edition exists or is sold, so it withholds nothing today.

Adoption

1 low confidence
1.0

Permit0 publishes no package on PyPI, npm or crates.io, so GitHub stars are the only signal. A star is not a use, so this shows attention rather than deployments.

Capability

3 medium confidence
3.0

Permit0 decides each agent action against policy before it runs, routes risky or unrecognized actions to a person, and signs its audit trail. It gives agents no identity of their own and has no way to pass narrowed authority from one agent or user to another.

Verified 2026-11-20, except some scores, still awaiting re-confirmation