AI Release Intelligence v0.1.0 is the first portfolio release of evidence-backed Go/No-Go decision support for GitHub-native releases.
Highlights
- Deterministic readiness assessment for one GitHub milestone and one release-candidate branch.
- Immutable evidence snapshots covering issues, pull requests, commits, CI checks, blockers, operational actions, migrations, and back-merges.
- Four explicit outcomes:
READY,NOT_READY,NEEDS_DECISION, andINSUFFICIENT_DATA. - Auditable human decisions for advisory risks, including reason, actor, timestamp, and evidence fingerprint.
- Stale decision invalidation when the underlying evidence changes.
- Optional AI explanations kept outside the readiness decision path.
- React, FastAPI, PostgreSQL, Docker Compose, Playwright, pytest, Vitest, Ruff, and mypy delivery stack.
- Public synthetic demo repository and a credential-free five-minute local walkthrough.
Try the MVP
Follow the five-minute deterministic demo to exercise the core flow without GitHub App, OAuth, or OpenAI credentials.
The public demo repository provides fictional GitHub evidence for the live integration scenario.
Verification
- CI run 33251685197 completed successfully on the tagged
maincommit. - The repository includes a versioned 44-scenario deterministic benchmark, PostgreSQL integration coverage, Docker smoke tests, and browser-level E2E coverage.
- Release commit:
7c5ea4f01b354c912a98b22fdbe0e6a355d30110.
Current scope
The MVP intentionally supports one repository, one milestone, and one candidate branch. Multi-repository releases, Jira and GitLab integrations, deployment orchestration, rollback, webhooks, scheduling, and notifications remain outside v0.1.0. External adoption and measured time savings are not yet validated.