v2.3 - Next.js frontend + JSON API split

The web app moves to a typed Next.js frontend calling a pure Flask JSON API.

• #17 fix(deploy): raise gunicorn's request timeout for /api/analyze
• #16 feat(web,frontend): migrate to Next.js + TypeScript frontend over a Flask JSON API

v2.2 - Web upload + premium landing

Resume file upload (PDF/DOCX) and a redesigned landing page.

• #14 feat(web): premium landing redesign + PDF/DOCX resume upload

(#13 and #15 covered the same ground and were closed unmerged in favor of #14.)

v2.1 - Grounded match fidelity

Replaces the token matcher with an LLM-graded, evidence-citing verdict (backed / not_shown / not_verifiable).

• #12 feat(analysis): ground the backed verdict in real repo code

v2.0 - Two faces on one engine

A no-install web app ships over the same core/ engine, alongside the MCP server.

• #11 style(web): apply ruff format to satisfy CI format check
• #10 docs(readme): add module fallback for running the web app
• #9 feat(web): add v2.0 no-install web app over the same core engine

v1.0 - Production polish

First shippable, public version.

• #8 docs: polish README and add write-up for the v1.0 release
• #7 chore(docker): add Dockerfile and verify the build in CI
• #6 feat(github): respect rate-limit signals with backoff and retry
• #5 chore(ci): add GitHub Actions running ruff, mypy, and pytest
• #4 fix(anthropic): stop truncating suggest_projects JSON responses

v0.3 - The prescription

suggest_projects() — the star tool: a ranked 30-day build plan tied to real resume claims.

• #3 feat(server): add suggest_projects 30-day build plan (v0.3)

v0.2 - The gap finder

analyze_resume() — extracts resume claims and cross-references them against real repos.

• #2 feat(server): add analyze_resume gap finder (v0.2)

v0.1 - Walking skeleton

MCP plumbing works end-to-end with one real tool.

• #1 feat(server): v0.1 walking skeleton with fetch_github_repos