MaintainerOps AI

Version updated for https://github.com/rtonf/maintainerops-ai to version v0.1.12.

  • This action is used across all versions by 0 repositories.

Action Type

This is a Composite action.

Go to the GitHub Marketplace to find the latest changes.

What’s Changed

MaintainerOps AI v0.1.12 expands model-backed eval readiness and records the first passing 10-case live eval run.

Highlights:

  • Added a dedicated 10-case model-backed eval file.
  • Added smoke / expanded / all suites plus targeted –case execution.
  • Added –cases-file, –summary-json, and API-free npm run eval:model:list.
  • Added stricter recommended-action and risk-bound checks.
  • Improved normalization for dependency updates, license metadata issues, and direct security-boundary findings.
  • Recorded the approved 10-case live model-backed eval result.

Live eval evidence:

  • Command: npm run eval:model – –suite all –budget-usd 0.5 –max-cases 10 –max-output-tokens 1200 –summary-json
  • Result: passed 10 cases.
  • Model: gpt-4o-mini.
  • Successful-run estimated cost: $0.001724.
  • Cumulative estimate across failed/adjusted/passing runs: $0.005114.
  • Approved ceiling: $0.50.

Guardrails remain unchanged: live model-backed evals are manual-only, not part of CI, and MaintainerOps AI does not auto-merge, auto-close, auto-label, publish releases, or scan unauthorized repositories.