crescerelabs

Sentience Governor

Name: Sentience Governor
Author: crescerelabs

Security community

SOURCE https://github.com/crescerelabs/sentience-governor

Description

Local-first runtime governance for AI agents. Declare intent, catch out-of-scope actions, and audit them in a verifiable trail.

Installation

This entry records only its repository, not the path inside it, so there is no exact command to give. Open the source below and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

README

Sentience Governor

**Local-first governance for AI agents at runtime.**

Sentience Governor captures agent actions at the execution boundary, evaluates them against declared intent, scope, and policy, and writes a verifiable local record of each captured tool call.

When an agent operates outside what it declared, Sentience surfaces the policy violation and attributes the associated token usage to the turn where it happened.

[![PyPI](https://img.shields.io/pypi/v/sentience-governor.svg)](https://pypi.org/project/sentience-governor/) [![Python](https://img.shields.io/pypi/pyversions/sentience-governor.svg)](https://pypi.org/project/sentience-governor/) [![License](https://img.shields.io/badge/license-Apache%202.0-blue.svg)](https://github.com/crescerelabs/sentience-governor/blob/main/LICENSE) [![test](https://github.com/crescerelabs/sentience-governor/actions/workflows/test.yml/badge.svg)](https://github.com/crescerelabs/sentience-governor/actions/workflows/test.yml)

No account, no API key. By default, your traces stay on your machine.

![Sentience Governor: an agent's declared intent on the left, its runtime actions on the right, each marked within scope, outside scope, or a policy violation, with token spend attributed to each action](https://raw.githubusercontent.com/crescerelabs/sentience-governor/main/docs/assets/demo.gif)

*Illustrative. Sentience records and evaluates a session, then reports on it; the open-source release does not adjudicate or block actions as they happen. For real output, see [See it work](#see-it-work) below.*

What's new

**0.3.1** adds `sentience scan` — a retrospective review of the Claude Code history already on your machine. It reports which sessions recorded write activity outside the project they were working in, before you have declared an intent or instrumented anything. Local and read-only: no signup, no network request, and no writes of any kind. The same review is available inside Claude Code as `/sentience-review`.