Skill: /implement-plan

Usage

/implement-plan

Invoke with no arguments. The plan from the previous /plan-feature session is the source of truth.

Prerequisites

• /plan-feature must have been approved by the lead in this session
• If the plan is not approved, do NOT proceed — re-run /plan-feature

Implementation Rules

The Golden Rules

1. Implement ONLY what the approved plan lists. No additions, no refactors, no "while I was in there" improvements.
2. Follow the agent's coding standards (defined in the agent prompt) at all times. Do not deviate for convenience.
3. Work in dependency order: database → models → repositories → services → controllers → routes → frontend → tests.
4. Stop and report if you discover a problem not in the plan — do NOT silently work around it.

Execution Process

1. Re-read the Approved Plan

Before writing a single line, read back the approved plan and output:

IMPLEMENTING PLAN:
──────────────────────────────────────────
Task    : {description}
Files   : {total count}
Order   : {numbered file list}
──────────────────────────────────────────

2. Implement File by File

For each file in the plan, follow this mini-cycle:

Step A — Pre-check If the file already exists, read it fully before modifying. Note the existing structure to ensure your changes integrate cleanly.

Step B — Implement Write or modify the file following the agent's coding standards.

Step C — Progress Report After completing each file, output:

✅  [{N}/{total}] {filename}
    → {1-line summary of what was done}

Step D — Dependency Pause (for multi-technology tasks) If the plan crosses a technology boundary (e.g., Laravel → React), pause and confirm:

⏸  Backend implementation complete. Ready to proceed to React layer.
   Confirm to continue, or flag any backend concerns first.

3. Handle Discoveries

If during implementation you discover:

A) The plan is incomplete (e.g., a required file wasn't listed):

🔎 DISCOVERY: {file} needs to be modified but was not in the plan.
   Proposed addition: {brief description}
   Awaiting lead confirmation before touching this file.

B) A technical blocker (e.g., the DB schema doesn't support the feature):

🚧 BLOCKER: {description of the problem}
   Cannot proceed with {step} until resolved.
   Options: {option A} | {option B}
   Awaiting lead decision.

C) A risk not caught in planning (e.g., a breaking change discovered):

⚠️  RISK DISCOVERED: {description}
   This was not flagged in the plan. Recommend pausing.
   Awaiting lead sign-off to continue.

4. Completion Report

When all files in the plan are implemented:

──────────────────────────────────────────────────────────────
  IMPLEMENTATION COMPLETE
──────────────────────────────────────────────────────────────
Files created : {count} — {list}
Files modified: {count} — {list}
Files deleted : {count} — {list}

Discoveries   : {count} — {list, or "none"}

Next step: /git-commit
──────────────────────────────────────────────────────────────

Do NOT run tests or linting at this stage — that happens in Step 7. Do NOT commit at this stage — that happens in Step 5.