Skip to content

README.md

Latest commit

 

History

History
103 lines (77 loc) · 3.12 KB

README.md

File metadata and controls

103 lines (77 loc) · 3.12 KB
Raw

OpSentry GitHub Action

Scan PRs for security vulnerabilities, code quality issues, and guardrail compliance.

Quick Start

Add to .github/workflows/opsentry.yml:

name: OpSentry
on:
  pull_request:
    branches: [main]

permissions:
  contents: write
  pull-requests: write

jobs:
  scan:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: opsight-intelligence/opsentry-action@v1

That's it. Every PR gets scanned for secrets, SQL injection, dangerous patterns, code quality, and guardrail compliance.

Inputs

Input Default Description
scan-type all security, code-health, governance, or all
config-path '' Path to a guardrails.yaml config file
scan-mode changed changed (PR files only) or full (all files)
fail-on high Not implemented yet -- accepted but ignored; see below
auto-fix true Auto-fix safe issues and commit to PR branch
post-comment true Post results as a PR comment
llm-provider none LLM for docstring generation: bedrock, anthropic, openai, local, none

Known limitations

  • fail-on is not implemented yet. The input is accepted but no step reads it, so the action does not fail the check based on finding severity. Do not rely on it as a merge gate.
  • Outputs are not populated yet. action.yml declares findings-count, critical-count, high-count and report-path, but none of them is set, so they are always empty. Do not reference them from later steps.

Outputs

Output Status
findings-count Declared, not populated
critical-count Declared, not populated
high-count Declared, not populated
report-path Declared, not populated

Examples

Security scan only

- uses: opsight-intelligence/opsentry-action@v1
  with:
    scan-type: security
    fail-on: critical

Full scan with LLM docstrings

- uses: opsight-intelligence/opsentry-action@v1
  with:
    scan-mode: full
    llm-provider: anthropic
  env:
    ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}

LLM provider credentials

The action does not take API keys as inputs. Pass whatever credentials your chosen llm-provider needs as env: on the step, sourced from repository secrets -- for example ANTHROPIC_API_KEY for anthropic, the provider's standard API-key variable for openai, or AWS credentials (for example via aws-actions/configure-aws-credentials) for bedrock. local and none need no credentials. Never put a key in the workflow file itself.

Versioning

This action follows Semantic Versioning. Each release is tagged vX.Y.Z, and the floating v1 tag always points at the newest v1.x.y release.

  • Pin to @v1 to receive backwards-compatible updates automatically
  • Pin to @v1.0.1 to lock to an exact release

See CHANGELOG.md for the release history.

Contributing

See CONTRIBUTING.md for the branching model, versioning rules, and what every change is expected to include.