wtf.implement-task（wtf 执行任务）

Implement Task

Pick up an existing Task as a developer. Core value: reads the full spec (Task + Feature + Epic), maps it to the actual codebase, proposes a concrete technical approach, then drives implementation test-first against each Gherkin scenario.

The expected Task issue body structure is defined in @.github/ISSUE_TEMPLATE/TASK.md.

Process

0. GitHub CLI setup

Run steps 1–2 of ../references/gh-setup.md (install check and auth check). Stop if gh is not installed or not authenticated. Extensions are not required for this skill.

Skip this step if invoked from wtf.verify-task or another skill that already ran gh-setup this session.

1. Identify the Task

Ask: "Which Task are you implementing? (issue number)"

Fetch the Task first, extract Feature and Epic numbers from its Context section, then fetch Feature and Epic in parallel:

gh issue view <task_number>    # Gherkin, Contracts, Impacted Areas — also yields feature and epic numbers
# Extract feature and epic numbers, then in parallel:
gh issue view <feature_number> # ACs, user stories
gh issue view <epic_number>    # Goal, context, constraints

2. Lifecycle check

Check whether the task has been designed:

gh issue view <task_number> --json labels --jq '.labels[].name'

If the designed label is absent, warn the user that the task hasn't been designed yet and that the recommended flow is: write-task → design-task → implement-task → verify-task. Then call AskUserQuestion with:

• question: "This task doesn't have a designed label yet. How would you like to proceed?"
• header: "Design check"
• options: [{label: "Design it first", description: "Go back and runwtf.design-task(default)"}, {label: "Skip design", description: "Proceed to implementation anyway"}]
• Design it first → follow the wtf.design-task process, passing the Task number in as context.
• Skip design → proceed.

If the designed label is present, continue silently.

3. Load the technical steering document

Use the Read tool to attempt reading docs/steering/TECH.md.

If the file exists: keep its content in context. Use its stack, architecture patterns, key constraints, commands, and ADRs to inform the technical approach and implementation in this session. Do not surface it to the user — just apply it silently.

If the file does not exist, call AskUserQuestion with:

• question: "docs/steering/TECH.md doesn't exist yet. This document captures your stack, architecture patterns, and technical constraints. Would you like to create it now?"
• header: "Tech steering doc missing"
• options: [{label: "Create it now", description: "Runwtf.steer-techbefore continuing (recommended)"}, {label: "Skip for this session", description: "Continue without it — technical decisions won't reference project standards"}]
• Create it now → follow the wtf.steer-tech process, then return to this skill and continue from step 4.
• Skip for this session → continue without it.

4. Set up the branch

Before writing any code, set up branches following the trunk-based feature branching strategy:

main
└── feature/<feature-number>-<feature-slug>   (merges → main)
    └── task/<task-number>-<task-slug>         (merges → feature branch)

Slug generation: For both the feature slug and task slug, spawn a subagent using the claude-haiku-4-5-20251001 model. Pass in the title and ask for a 2–4 word kebab-case summary restricted to [a-z0-9-] characters (e.g. date-range-filter).

Feature branch — create if missing:

git fetch origin
git checkout feature/<feature-number>-<feature-slug> 2>/dev/null || {
  git checkout main
  git pull --rebase origin main
  git checkout -b feature/<feature-number>-<feature-slug>
  git push -u origin feature/<feature-number>-<feature-slug>
}
git pull --rebase origin feature/<feature-number>-<feature-slug>

Task branch — create or resume:

# Fresh work:
git checkout -b task/<task-number>-<task-slug>

# Resumed work (branch already exists):
git checkout task/<task-number>-<task-slug>
git rebase origin/feature/<feature-number>-<feature-slug>

Resolve any conflicts before proceeding. Print the branch name.

5. Explore the codebase

Before exploring, identify the test framework setup by reading a sample of existing test files. Record the following in a working scratchpad before proceeding — these govern every test written in step 8:

Use the Agent tool with these concrete searches (run in parallel):

• Grep for the domain nouns and verbs from the Task's Functional Description across *.{ts,tsx,js,jsx,py,go,rb} files — finds files and modules this task will touch
• Glob matching the file patterns for each Impacted Area listed in the Task (e.g. src/api/**/*, src/features/<feature-slug>/**/*) — surfaces integration points and existing patterns
• Grep for interface or type names from the Task's Contracts section — finds current interface definitions to implement against
• Glob matching the test file pattern from the scratchpad (e.g. **/*.test.ts) near the integration points found above — surfaces existing tests covering adjacent behavior
• Grep for any import of the domain objects or services this task depends on — identifies dependencies that must exist first

Also fetch any relevant wiki pages or in-repo glossary docs for this task's Bounded Context. Use these to ensure the implementation and test naming aligns with the team's Ubiquitous Language.

6. Draft the Technical Approach

Produce a concrete Technical Approach with actual file paths (not generic layer names):

• Architecture decisions: which layer owns what, which patterns to follow
• Data flow: how data moves from input to output
• Trade-offs: what alternatives were considered and why this approach was chosen
• Impacted Areas: concrete file paths for Backend, Frontend, Database, APIs

7. Review approach with user

Show the Technical Approach. Then call AskUserQuestion with question: "Does this align with how you'd approach it?", header: "Approach review", and options: [{label: "Yes — looks good, proceed", description: "Continue with implementation"}, {label: "I have constraints to share", description: "I want to adjust the approach first"}, {label: "Suggest an alternative", description: "Let me describe a different approach"}].

Apply changes. Then update the Task issue with the Technical Approach and Impacted Areas.

See references/issue-body-update-pattern.md for the read-merge-write pattern. Use /tmp/wtf.implement-task-<task_number>-approach.md as the temp file.

gh issue edit <task_number> --body-file /tmp/wtf.implement-task-<task_number>-approach.md

8. Drive the TDD cycle

For each Gherkin scenario in the Task, work through them in order. Match the project's established test patterns discovered in step 5. Reference the Contracts & Interfaces section for exact request/response shapes.

1. Write the failing test for the scenario.
2. Implement the minimum code to make it pass.
3. Refactor if needed — keep functions under 40 lines, no deep nesting.
4. Commit — atomic semantic commit per ../references/commit-conventions.md. Use the Scenario: and Task: trailers: git add <changed files> git commit -m "<type>(<scope>): <short description> Scenario: <scenario name> Task: #<task_number>"
5. Do not skip ahead — each scenario is a checkpoint.

Once all scenarios are green, run the full lint and type-check gate once across all changes. Check package.json for lint, typecheck, type-check, or check script keys and run whichever exist:

# e.g. npm run lint && npm run typecheck

Fix any issues before proceeding to coverage.

9. Verify coverage

Once all scenarios pass, confirm unit test coverage meets the minimum threshold for all new and modified code. Use the threshold specified in docs/steering/QA.md if it exists; default to 80% if the document is absent or does not define a threshold:

# Run the project's coverage command (check package.json scripts)

If coverage is below 80% on any new or modified file, add targeted tests before proceeding. Every public function must have at least one happy-path and one error-path test.

10. Update Test Mapping

Fill the Test Mapping table in the Task issue with concrete file paths:

See references/issue-body-update-pattern.md for the read-merge-write pattern. Re-fetch the body (do not reuse the temp file from step 6). Use /tmp/wtf.implement-task-<task_number>-test-mapping.md as the temp file.

gh issue edit <task_number> --body-file /tmp/wtf.implement-task-<task_number>-test-mapping.md

Print the updated Task issue URL.

11. Mark implemented and offer to continue

Add the implemented lifecycle label — this is mandatory regardless of invocation mode:

gh issue edit <task_number> --add-label "implemented"

If invoked from the loop (non-interactive mode), skip the AskUserQuestion below and return control to the loop.

Call AskUserQuestion with:

• question: "What's next?"
• header: "Next step"
• options: [{label: "Verify this Task", description: "Run QA against the Gherkin scenarios (recommended next step, default)"}, {label: "Open a pull request", description: "Create a PR for this branch"}, {label: "Implement another Task", description: "Implement another Task for the same Feature"}]
• Verify this Task → follow the wtf.verify-task process, passing the Task number in as context so the user is not asked for it again.
• Open a pull request → follow the wtf.create-pr process, passing the Task number and branch in as context.
• Implement another Task → restart this skill from step 1.