Sentience Governor — Documentation
See exactly what your AI agents did, why they did it, and where they crossed boundaries — in real time.
Installs from PyPI with one command. No account required.
Start here (2 minutes)
- Install →
pip install sentience-governor - Run → Quickstart
- See your first trace
Everything else can wait.
Definitions
Session = one run of an agent Event = one action inside a session Trace = all events for a session
Where to go
- Install — get the CLI on your machine.
- Quickstart — from zero to your first trace in three minutes.
- Commands — reference for every CLI command.
- Sync privacy — what Sentience Sync sends, and what it does not.
- Troubleshooting — common issues and their fixes.
- Changelog — what changed, and when.
CLI tools you get
sentience→ view agent sessions (start here)sentience-cli→ raw trace viewersentience-sync→ optional cloud syncsentience-claude-code-hook→ auto hook (Claude Code)
A governance runtime
Wraps your agent with five control points: AGENT_REGISTERED, INTENT_DECLARED, SCOPE_ASSERTED, CONTEXT_SNAPSHOT, MEMORY_WRITE_ATTEMPT.
Emits a structured event at each control point. Each event is evaluated against five default policy rules; violations are annotated with the consequence the paid control plane would have applied.
Nothing gets blocked. Everything is visible.
Trace files
Per-session traces land at ~/.sentience/traces/. You can read, grep, diff, replay, or feed them into your own analysis.
Sync CLI
Send aggregated counts — not raw events — to Sentience Cloud. Manual, explicit, opt-in.
One run at a time. Never automatic.
The philosophy
- Observe-only.
- Fail-open.
- No automatic network calls.
Who it is for
- Operators running production AI agents who need an audit trail.
- Teams adopting MCP or LangChain who need structured logs of tool-call behaviour.
- Claude Code users who want a record of every editor tool invocation with four lines of config.
- Developers evaluating governance who want to see what the open-tier surface looks like.
What the open tier does not do
Real enforcement — block, pause, scope contraction, retroactive revocation — is a paid control-plane capability. The open tier is the observability half of the same architecture.
The upgrade path is additive. Adopting open-tier today puts you on the path to full enforcement later.
Stuck? → Troubleshooting
Testing a pre-release or early-access build? Install instructions are here.