vibe-pm-kit

The product management kit for vibe coding. Specs an AI can build, prompt briefs, eval scorecards, and honest playbooks for training and tuning models, plus a zero-dependency spec linter that scores your spec before you build from it.

Live site: https://aamirtechs.github.io/vibe-pm-kit/ with the spec scorer running in your browser. Paste a spec, get the score, fix the gaps.

Vibe coding made building fast. It did not make deciding what to build, proving the model behaves, or knowing when to fine-tune any easier. This kit is the missing product layer: small, practical, and opinionated.

Quickstart

Score the included spec template (it scores 100):

python tools/spec_lint.py templates/spec-template.md --check

Then score your own spec before your next build session:

python tools/spec_lint.py path/to/your-spec.md

Example output:

Spec score: 55/100 for my-spec.md
  ok       Problem
  ok       Target users
  ok       Goals and scope
  ok       Workflows
  ok       Risks
  MISSING  Non-goals
  MISSING  Acceptance criteria
  MISSING  Data and models
  MISSING  Success metrics

What's inside

Playbooks (docs/)

• The vibe PM loop: frame, spec, brief, build, eval, ship, learn
• Writing specs an AI can build: the rules and the anti-patterns
• Training vs tuning: the decision ladder from prompts to RAG to fine-tuning to training, with the honest costs of each rung
• Evals: building eval sets and gating releases on them
• Launch and grow: shipping and spreading a vibe-built product without spam tricks

Templates (templates/)

• PRD template, spec template, prompt brief template
• Dataset card template (fill before any training or tuning run)
• Eval scorecard template, launch checklist

Tool (tools/)

• spec_lint.py: zero-dependency Python CLI that checks a Markdown spec for the nine sections an AI builder needs and scores it out of 100, with --check and --min-score for CI

Who this is for

• Solo founders and indie hackers building with AI coding agents
• Product managers whose team now ships with models in the loop
• Anyone who has watched a vibe-coded demo collapse on real data and wants the process that prevents it

Tests

python -m pytest tests/ -q
# or
python -m unittest discover -s tests

Contributing

Issues and PRs are welcome. GOOD_FIRST_ISSUES.md lists small, well-scoped starting points. Keep additions practical: if a practising PM cannot use it this week, it does not belong in the kit yet.

License

MIT. Copyright 2026 Amir Hussain.