Executing Plans

Load plan, review critically, execute all tasks with verification at each step, finish with adr-commit.

Prefer `subagent-driven-development` if your harness supports subagents. Executing-plans is for environments where subagent dispatch is not available.

Announce: "Using executing-plans to implement this plan."

Step 1: Load and Review Plan

1. Read plan file from docs/superskills/plans/
2. Review critically — identify questions or concerns
3. If concerns → raise with human before starting
4. If no concerns → create TodoWrite and proceed

Step 2: Execute Tasks

For each task:

1. Mark as in_progress in TodoWrite
2. Follow each step exactly — plan has exact code and commands
3. For each code step: use tdd-verified (write test → watch fail → implement → watch pass)
4. Run verifications as specified
5. After confirmed bug fix → trigger pattern-globalize
6. Before marking done → run verification-before-completion
7. Mark as completed

Step 3: Finish with adr-commit

After all tasks complete:

1. Announce: "Using adr-commit to complete this work."
2. Invoke adr-commit — ADR Gate + G.CDOC + PATTERN-DEBT lifecycle + merge options

Do NOT use finishing-a-development-branch. Use adr-commit.

When to Stop

Stop immediately when:

• Blocker that prevents continuing (missing dependency, repeated test failure)
• Plan has critical gaps preventing starting
• Instruction unclear
• Verification fails repeatedly

Ask for clarification. Do not guess through blockers.

Key Rules

• Review plan critically before starting
• Follow plan steps exactly — do not improvise
• Do not skip verifications
• Stop when blocked, ask for direction
• Never start on main/master without explicit user consent
• tdd-verified for every implementation step
• pattern-globalize after every confirmed bug
• verification-before-completion before every "task complete" mark

Gotchas — Observed Failure Points

<!-- Populated by knowledge-compound skill after cycles where this skill underperformed --> <!-- Format: [YYYY-MM-DD] What failed | Root cause | What to do instead --> <!-- DO NOT pre-populate with speculation — real observations only --> <!-- Example: [2026-06-01] Agent skipped risk scoring for "trivial" CSS task that touched auth config | Triage gate not read carefully | Triage gate now checks file content, not just task description -->