Coder

Implementation workflow

1. Understand request scope and inspect existing code paths.
2. Implement minimal diffs in the right layer (domain/template/shared).
3. Preserve existing naming and data boundary conventions.
4. Validate with REPL eval and/or focused tests.
5. Summarize changed files, checks run, and remaining risks.

Project-specific coding expectations

• Respect snake_case (DB) vs kebab-case (app/runtime) boundaries.
• For re-frame handlers with trim-v interceptors, destructure event args as [params].
• For backend JSON responses, convert PG-specific objects before encoding.
• Preserve existing session behavior and route boundaries unless change is requested.
• For UI interaction work, keep stable and unique component :id attributes.

Validation requirements

• Required for behavior changes and non-trivial edits:
  • At least one of: REPL validation or focused tests.
  • Cover edge cases when applicable: happy path, nil, empty, invalid/boundary input.
• If running tests from shell and saving output, save once to tmp/ and reuse it for analysis.

Forbidden shortcuts

• Do not rewrite large files when a small targeted change solves the issue.
• Do not invent schema details; inspect via postgres-mcp.
• Do not bypass migration workflow for schema updates.