AnxLabs Sentinel
Local-first security guardrails for coding-agent workflows.
The project
Checks before the command runs.
Sentinel is a local-first defensive security guardrail for AnxLabs coding-agent workflows, implemented as an independent ZCode plugin.
Deterministic pre-execution hooks classify every guarded operation as SAFE, REVIEW, or BLOCK — covering command execution, file writes, secrets, sensitive paths, and git history. Version 0.1 performs its checks locally, with no network calls and no telemetry.
What it explores
- Command and git guardrails with SAFE, REVIEW, and BLOCK verdicts
- Secret and sensitive-path checks on file writes
- Local-only configuration and redacted logging
- Evidence-based security review command
Where it stands
Version 0.1, implemented and validated on Windows. A public repository or download has not been confirmed, so none is offered here.
In the ecosystem
Security tooling for the coding-agent workflows used around AnxLabs. It runs independently of other projects; no integration is claimed.
Keep exploring