F2: GitHub Actions Generation Engine — MVP

## Summary
AI generates a tailored GitHub Actions workflow on first successful deploy. All subsequent deploys run through that fixed, versioned Action — making repeat deploys deterministic.

## Acceptance Criteria
- [ ] On first successful deploy, AI authors a GitHub Actions workflow tailored to the repo
- [ ] Top frameworks supported at MVP: Node.js, Python
- [ ] Subsequent deploys run through the generated Action (not re-analyzed by AI)
- [ ] AI's ongoing role narrows to authoring/adjusting the Action file
- [ ] Clear in-progress feedback during first-time analysis/build
- [ ] Build caching and pre-warmed base images for supported frameworks

## Sprint Assignment
- Sprint 1–2: Design & build
- Sprint 3: MVP ships (Node.js, Python)
- Sprint 4–5: Framework expansion (background)

## Success Metrics
- Median time-to-first-deploy
- Deploy determinism: zero variance on repeated deploys of unchanged code
- Framework coverage rate

No files, tests, or entry points are named. Start by locating the existing successful-deploy flow and any GitHub Actions or framework-detection integration, then trace how deploys are currently analyzed. Done means a first successful Node.js or Python deploy generates a versioned workflow, later unchanged deploys reuse it deterministically, and users receive progress feedback.

Written by the indexing model from the issue text.