codify

codify — Stage 3 of the Spark lifecycle

Ideate → Plan → Generate → Solve → Ship

The codify skill implements exactly one issue. Its job is to turn acceptance criteria into working code on a branch — nothing wider. It owns the coding lane (Generate, stage 3).

Do this

1. Read the issue. The acceptance criteria are the contract. If they're missing or vague, go back to plan — don't guess.
2. Branch. Never work on master/main. Create a focused branch:

   git checkout -b feat/<short-slug>      # or fix/, refactor/, docs/
   

3. Match the surrounding code. Read neighboring files first; mirror their naming, structure, and idioms. Write code that reads like it was already there.
4. Implement to the criteria, then stop. Resist scope creep — anything not in the issue is a new issue, not a freebie.
5. Self-check against each acceptance criterion before declaring done.

Guardrails

• One issue per branch, one concern per branch.
• No commented-out code. No debug prints in library code.
• Don't add dependencies the issue doesn't require.
• Don't refactor surrounding code opportunistically — open a separate issue.
• Follow the project's CLAUDE.md standards (types, docstrings, formatter, linter) — run them before handing off.
• Don't write documentation — that's docit (outward) or knowledge (internal). codify writes code.

Next

Send the change to fix-issue to review and harden, then to ship to commit and open a PR.