skills/core/executing-plans/SKILL.md
Use when implementing a multi-step plan. Execute systematically with verification checkpoints to catch errors early.
npx skillsauth add liauw-media/codeassist executing-plansInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Execute plans systematically, one task at a time, with verification checkpoints. NEVER work through multiple tasks without verification.
writing-plans skill and getting user approvalExecute ONE task at a time. Verify EACH task before proceeding.
Reasons:
Template:
I'm using the executing-plans skill to implement this systematically.
I'll work through each task one at a time, verifying as I go.
BEFORE starting ANY work:
Marking Task 1 as in_progress.
Update TodoWrite status to in_progress.
CRITICAL: Only ONE task should ever be in_progress at a time.
Follow the task's action items EXACTLY as written in the plan:
MANDATORY after each task:
Template:
Verifying Task 1:
- Running: [verification command]
- Expected: [expected outcome]
- Actual: [actual outcome]
- Status: ✅ PASS / ❌ FAIL
ONLY after successful verification:
Task 1 completed successfully. Moving to Task 2.
Update TodoWrite status to completed.
Before moving on, explicitly state:
Marking Task 2 as in_progress.
NEVER start a new task without:
If a task fails verification:
Template:
❌ Task [X] verification FAILED
Expected: [what should have happened]
Actual: [what actually happened]
Issue: [description of problem]
I'm investigating before proceeding. Keeping Task [X] as in_progress.
Actions to take:
in_progress (don't mark completed)If issue is blocking:
This issue is blocking progress. I need to resolve it before continuing.
Options:
1. [Fix approach 1]
2. [Fix approach 2]
Which approach would you prefer?
For long-running plans, provide status updates:
Template:
Progress Update:
✅ Task 1: [Name] - Completed
✅ Task 2: [Name] - Completed
🔄 Task 3: [Name] - In Progress
⏳ Task 4: [Name] - Pending
⏳ Task 5: [Name] - Pending
Currently working on: [Current task description]
MANDATORY: Use database-backup skill
Before proceeding with Task [X] (database operation), I'm using the database-backup skill.
MANDATORY: Use database safety wrapper
Running tests with safety wrapper: ./scripts/safe-test.sh [test command]
in_progress → Only ONE at a timeIf you discover issues during execution:
Template:
During Task [X], I discovered [issue/new information].
This affects the plan:
- [Impact 1]
- [Impact 2]
I recommend [adjustment to plan].
Should I:
1. Continue with original plan
2. Adjust plan as suggested
3. Stop and re-plan
NEVER silently deviate from the plan.
Before executing-plans: Use writing-plans skill
During execution: Use database-backup, test-driven-development as needed
After execution: Use code-review and verification-before-completion skills
I'm using the executing-plans skill to implement user authentication.
Plan has 6 tasks. I'll execute them one at a time with verification.
Marking Task 1 as in_progress: Install and Configure Sanctum
Task 1: Install and Configure Sanctum
Actions:
1. Running: composer require laravel/sanctum
✅ Package installed
2. Publishing config: php artisan vendor:publish --provider="Laravel\Sanctum\SanctumServiceProvider"
✅ Config published
3. Database backup (MANDATORY before migration)
I'm using the database-backup skill.
✅ Backup created: backups/database_2025-01-06_14-30-00.sql
4. Running migration: php artisan migrate
✅ Migrations completed
5. Adding Sanctum middleware to app/Http/Kernel.php
✅ Middleware added
Verifying Task 1:
- Running: php artisan route:list | grep sanctum
- Expected: Sanctum routes should be visible
- Actual: Found sanctum routes in output
- Status: ✅ PASS
Task 1 completed successfully.
Marking Task 2 as in_progress: Create Authentication Controller
Task 2: Create Authentication Controller
Actions:
1. Creating app/Http/Controllers/AuthController.php
✅ File created
2. Implementing register() method with validation
✅ Method added
3. Implementing login() method with token generation
✅ Method added
4. Implementing logout() method with token revocation
✅ Method added
Verifying Task 2:
- Running: cat app/Http/Controllers/AuthController.php | grep "public function"
- Expected: Should see register, login, logout methods
- Actual: All three methods present
- Status: ✅ PASS
Task 2 completed successfully.
Marking Task 5 as in_progress: Run Tests with Database Backup
Task 5: Run Tests with Database Backup
Actions:
1. Using database-backup skill
✅ Backup created: backups/database_2025-01-06_14-45-00.sql
2. Running: ./scripts/safe-test.sh vendor/bin/paratest --filter=AuthenticationTest
❌ Test failed
Verifying Task 5:
- Running: vendor/bin/paratest --filter=AuthenticationTest
- Expected: All authentication tests pass
- Actual: 2 failures in login test
- Status: ❌ FAIL
Error output:
Failed asserting that 401 matches expected 200. Expected token in response, got error message.
Issue: Login endpoint returning 401 instead of 200. Appears to be credentials validation issue.
Investigating AuthController login method...
Found issue: Missing password verification.
Fixing: Added Hash::check() for password validation.
Re-running verification:
- Running: ./scripts/safe-test.sh vendor/bin/paratest --filter=AuthenticationTest
- Status: ✅ PASS - All tests now passing
Task 5 completed successfully (after fix).
Progress Update - User Authentication Implementation:
✅ Task 1: Install and Configure Sanctum - Completed
✅ Task 2: Create Authentication Controller - Completed
✅ Task 3: Add API Routes - Completed
✅ Task 4: Create Authentication Tests - Completed
✅ Task 5: Run Tests with Database Backup - Completed (fixed login validation)
🔄 Task 6: Update API Documentation - In Progress
Currently working on: Generating OpenAPI documentation for auth endpoints
Task 6: Update API Documentation
Actions:
1. Adding PHPDoc comments to AuthController methods
✅ Comments added
2. Generating OpenAPI docs: php artisan scramble:generate
✅ Documentation generated
3. Verifying endpoints in /docs/api
✅ Auth endpoints visible
Verifying Task 6:
- Running: curl http://localhost/docs/api | grep "auth"
- Expected: Should see register, login, logout endpoints
- Actual: All three endpoints present in documentation
- Status: ✅ PASS
Task 6 completed successfully.
---
All 6 tasks completed successfully! ✅
Summary:
✅ Sanctum installed and configured
✅ Authentication controller created
✅ API routes exposed
✅ Tests created and passing
✅ API documentation generated
Ready for code-review skill.
Before marking a task as completed:
This skill is based on:
Social Proof: Professional developers verify each step. No one writes 1000 lines without running tests.
Before using this skill, confirm:
Bottom Line: Slow and steady wins. One task, one verification, one completion at a time. Batch work leads to batch failures.
development
Use when decomposing complex work. Dispatch fresh subagent per task, review between tasks. Flow: Load plan → Dispatch task → Review output → Apply feedback → Mark complete → Next task. No skipping reviews, no parallel dispatch.
development
# Server Documentation System Set up a documentation system that tracks changes and maintains server/project documentation with Claude Code hooks. ## When to Use - Setting up a new server or development environment - Need to track configuration changes over time - Want automatic documentation of work sessions - Maintaining changelog for infrastructure ## Directory Structure ``` ~/docs/ # User home directory (cross-platform) ├── changelog.md # Global over
development
Delegate tasks to remote Claude Code agent containers for parallel execution, long-running analysis, or resource-intensive operations.
development
Use when working on multiple features simultaneously. Creates isolated workspaces without branch switching, enabling parallel development.