home/career/execute-plan-prompt

Execute Plan.prompt

GPTClaudeGemini··1,148 copies·updated 2026-07-14
execute-plan-prompt.prompt
---
description: Execute an implementation plan — implement, test, document, and close out
agent: agent
---

# Execute Implementation Plan

You have been given a plan file. Your job is to **implement it fully** — code, test, document, and close out.

## Instructions

### 1. Read the Plan

- Read the attached plan file completely.
- Understand every section: Context, Current State, Target State, Assumptions, Implementation Steps, Testing Plan, Post-Implementation Checklist.
- Note the gate structure.

### 2. Review Recent Context

- **Read latest task log(s)** in `docs/task/logs/`.
- **Read `docs/task/todo.md`** for conflicts.
- **Read `docs/task/lessons.md`** for relevant lessons.
- **Read agent observation logs** in `docs/agent-observations/` for relevant unresolved items.

If the plan is stale or conflicts, **STOP and tell the user**.

### 3. Execute Implementation Steps

Work through each step in order:

- Mark checkboxes as you complete them.
- Follow exact file paths, line numbers, and descriptions.
- If line numbers have drifted, locate by context.
- If blocked or plan is wrong, **STOP and tell the user**.

### 4. Execute Testing Plan (BLOCKING)

After ALL implementation steps, execute the Testing Plan:

1. Run every Automated Check (`composer test`, `composer lint`).
2. Execute every Browser Verification step using Chrome MCP tools. Follow `devtools.instructions.md`.
3. Execute every Regression Check via browser.
4. Verify Edge Cases.

**Reading code is not testing. You must use Chrome MCP tools.** If any test fails, fix and re-test.

### 5. Documentation & Logging (only after testing passes)

1. Update affected documentation per `doc-sync.instructions.md`.
2. Add task log entry to today's `docs/task/logs/YYYY-MM-DD.md`.
3. Log observations per `agent-observations.instructions.md`. If none: "Observations: none."

### 6. Close Out

1. Update plan status to **Complete**.
2. Mark all Post-Implementation Checklist items checked.
3. Remove task from `docs/task/todo.md`.
4. Move plan file from `draft/` to `completed/`.
5. Confirm to user: what was done, what was tested, result.

## Rules

- **Follow the plan exactly.** No adding/skipping/reordering unless blocked.
- **Testing is not optional.** Gate 2 must pass before Gate 3.
- **Do not ask the user to test.** You are the tester.
- **If the plan is outdated**, STOP and tell the user.
- **Do not close with undocumented observations.**
- **No backup files in source directories.**
- **Follow task.instructions.md** for the gate-checked pipeline.

when to use it

Community prompt sourced from the open-source GitHub repo savedpixel/ai-agent-rules-generator (MIT). A "Execute Plan.prompt" style prompt — adapt the placeholders and specifics to your task. Imported as-is and not independently retested here, so check the output before relying on it.

tags

careercommunitygeneral

source

savedpixel/ai-agent-rules-generator · MIT