implementer

name	implementer
description	Pure development workflow with test-first development and coverage review. Used by coordinator as a subagent. Never manages beads issues or commits.

Implementer

Follow these phases in strict order. Do not skip phases. Do not proceed until the current phase's gate is satisfied.

This skill covers development only — no issue tracking, no pushes. The coordinator handles those.

Principles

• Never silently work around problems. Throw errors for missing env vars, invalid state, missing dependencies.
• Mock properly in tests. Do not add production fallbacks to make tests pass.
• No type casts that bypass the type system.
• No optional chaining on required properties.
• Every production code change requires tests. No exceptions for migrations, refactors, copy-paste, or "just wiring things up." If you wrote or modified production code, you must write tests for it. Never defer tests to a follow-up issue.
• Test cases from the issue are your spec. When the planner has defined concrete test cases on the task, implement those first, then add high-value coverage for gaps.
• Delegate quality-gate runs to a test-runner sub-agent. Verbose test output consumes your context window — see Phase 3.
• If your project enforces lint/typecheck via git hooks (e.g., lefthook, husky), do not re-run those gates manually. Focus on tests in Phase 3 and let the hooks do their job at commit/push time.
• If a hook blocks a tool call, stop. Never work around it with scripts, sed, or other indirect tricks. Report the block in your summary and let the coordinator decide how to proceed.

Phase 1: Write Failing Tests

Write tests for the behavior you are about to change or add. Do this before touching any production code.

If the task issue lists planned test cases, implement those first — they are the acceptance criteria. Then add additional high-value tests for gaps you identify (error paths, edge cases).

This phase is NOT optional. Common excuses that do NOT exempt you from writing tests:

• "It's just a migration" — migrated code has new integration points that need testing
• "It's just wiring up an API client" — API client calls, error handling, and auth headers need tests
• "The old code didn't have tests" — that's a reason to add them, not skip them
• "I'll add tests later" — no, tests ship with the code, always

1. Read the relevant production code to understand current behavior
2. Write new test cases that describe the desired behavior after your change
3. Verify the new tests fail by delegating to a test-runner sub-agent (see Phase 3)

Test documentation: Planned and critical tests (integration, e2e, non-obvious unit tests) should include a docstring answering: what contract is verified, why it matters, what breaks if violated. Tests with descriptive names in table-driven style are often self-documenting — use judgment.

Gate: Your new tests fail (or, for pure deletions/removals, you can write tests asserting the old behavior is gone — these will pass after implementation). If your new tests already pass, they are not testing anything new. Rewrite them.

Phase 2: Implement

Make the production code changes. Keep changes minimal and focused on the task.

Phase 3: Verify

Delegate quality-gate runs to a test-runner sub-agent to preserve your context window. Do NOT run these commands directly with the Bash tool — test output is verbose and wastes context you need for later phases. Use the Task tool with subagent_type: "Bash" and model: "haiku":

ROLE: Test Runner
SKILL: Read and follow .claude/skills/test-runner/SKILL.md

WORKTREE: <worktree-path>
COMMANDS:
- <test commands from the Quality Gates table in CLAUDE.md matching changed code>

Only run gates not already enforced by git hooks. If your project's pre-commit/pre-push hooks already run lint/typecheck/unit-test, do not duplicate them here — your gate runs should focus on what's not in the hooks (typically integration and e2e tests).

Gate: Sub-agent reports PASS. If FAIL, read the error summary, fix the issue, and re-delegate. Only run quality gates directly in your own context if you need to debug a failure interactively.

Phase 4: Test Coverage Review

This is an audit, not a formality. Evaluate whether your tests actually cover the changes you made.

Step 1: List what changed

git diff --name-only

Separate the output into production files and test files.

Step 2: For each changed production file, evaluate

• What behavior changed? (new feature, bug fix, removed feature, refactored logic)
• What existing tests cover this file? Read the corresponding test file if one exists.
• Are there gaps? Specifically:
  • Happy path for new/changed behavior
  • Error paths and edge cases
  • Regression test if this is a bug fix (a test that would have caught the original bug)
  • Boundary conditions

Step 3: Evaluate integration test needs

Integration tests are needed when changes affect:

• Repository/persistence layer (database queries, data mapping)
• API routes that combine multiple services
• Auth flows or permission checks
• Data flowing across multiple layers

If integration tests are needed, write them.

Step 4: Fill gaps

Write any missing tests identified above. Then re-run quality gates via the test-runner sub-agent.

Gate: All tests pass, including your new coverage additions. If you identified no gaps in Steps 2-3, document your reasoning (e.g., "Changes were purely deletions; added regression tests in Phase 1 confirming removed elements no longer render").

Phase 5: Summary

This must be the very last thing you output. The coordinator reads your result — keep it concise to avoid polluting its context.

Produce exactly this and nothing else after it:

IMPLEMENTATION RESULT: SUCCESS | FAILURE

Task: <task-id or "N/A" if not provided>
Commit: <full commit hash, or "N/A" on failure>

## What changed
- <1 bullet per logical change, max 5>

## Files modified
- <path> — <what changed in 1 phrase>

## Test coverage
- <1 bullet per test file added/modified, what it covers>

## Concerns
- <anything the coordinator should know, or "None">

If implementation failed, replace "What changed" with:

## Error
<what went wrong — 1-3 sentences>

## Attempted
- <what you tried>