# Trustabl

Trustabl is a GitHub Action that scans an agent codebase across frameworks such as the Claude Agent SDK, OpenAI Agents SDK, Google ADK, MCP, LangChain, LangGraph, CrewAI, AutoGen and Pydantic AI, and reports production reliability findings directly in the pull request.

The action downloads a signature-verified release binary of the trustabl scanner, runs it against the checkout, and uploads results as inline annotations on the changed lines of a pull request, entries in GitHub's Security tab, and a sticky pull request comment with a readiness score, a severity breakdown and suggested fixes. It can also gate a pull request by failing the job when a configurable risk-score or severity threshold is crossed, and can optionally match declared dependencies against a pinned vulnerability snapshot to report known CVEs alongside its own findings.

With no configuration beyond adding the action, it scans the checkout on every push and pull request and fails the job only when the scanner itself flags a medium-or-higher finding; missing permissions degrade gracefully to whichever reporting surfaces remain available rather than failing the run.

It is written in TypeScript and released under Apache-2.0.

Website: https://github.com/trustabl/trustabl-action