Os Large

---
description: Full autonomous implementation of a GitHub issue from planning to review
argument-hint: <issue-number>
allowed-tools:
  - Task
  - TodoWrite
  - Bash
  - Read
  - Write
  - Edit
  - Glob
  - Grep
---

# One Shot Large: $ARGUMENT

## Initialize TodoWrite Progress Tracking
Use TodoWrite to create this checklist:

• Retrieve GitHub issue and setup branch
• Generate technical specification and tasks
• Parallel implementation (engineering, testing, infrastructure)
• Run all tests and quality checks
• Code review and final verification

## PHASE 1: Setup
*Mark "Retrieve GitHub issue and setup branch" as in_progress*

### Step 1.1: GitHub Issue and Git Setup

EXECUTE:

1. Fetch issue: gh issue view $ARGUMENT --json number,title,body,state,assignees,author,labels,createdAt,updatedAt
2. git stash
3. git status
4. git checkout master
5. git pull
6. Derive branch name: TITLE=$(gh issue view $ARGUMENT --json title -q .title) CLEAN=$(echo "$TITLE" | tr '[:upper:]' '[:lower:]' | tr -cs '[:alnum:]' '-' | cut -c1-50) BRANCH="feature/$ARGUMENT-${CLEAN}"
7. git checkout -b $BRANCH
8. mkdir -p docs/tickets/$ARGUMENT docs/tickets/$ARGUMENT/attachments

ACTIONS:

• Fetch issue body: gh issue view $ARGUMENT --json body -q .body
• Check for attachments (images, specification files) in markdown
• Write GitHub issue content to docs/tickets/$ARGUMENT/ticket.md (title, state, body)
• If specification files found in body:
  • Download using curl
  • Append to ticket.md with separator
  • Delete downloaded specification files VERIFY: Branch created, folder exists, issue documented

*Mark "Retrieve GitHub issue and setup branch" as completed*

## PHASE 2: Specification Development
*Mark "Generate technical specification and tasks" as in_progress*

### Step 2.1: Create Specification

STEP 1: Create Technical Specification AGENT: full-stack-engineer TASK: Create technical specification PROMPT FOR AGENT: "Create docs/tickets/$ARGUMENT/specification.md based on docs/tickets/$ARGUMENT/ticket.md covering:

1. High level technical implementation approach
2. File structure and routing (use libs/ per CLAUDE.md)
3. Error handling implementation
4. RESTful API endpoints if needed
5. Database schema if needed IMPORTANT: Only focus on this issue, no cross-cutting concerns."

STEP 2: Create Infrastructure Assessment AGENT: infrastructure-engineer TASK: Assess infrastructure needs PROMPT FOR AGENT: "Review docs/tickets/$ARGUMENT/specification.md and determine infrastructure needs. If changes needed, ADD infrastructure section to docs/tickets/$ARGUMENT/specification.md covering:

1. Database changes
2. Environment variables
3. Helm chart updates
4. Docker/Kubernetes updates
5. CI/CD pipeline changes If no changes needed, skip this step. IMPORTANT: Only focus on this issue."

STEP 3: CREATE TASK LIST TASK: Create task breakdown ACTION: "Create docs/tickets/$ARGUMENT/tasks.md with structure:

Implementation Tasks (full-stack-engineer)

• [List implementation tasks from specification]
• Write unit tests for all new code

Testing Tasks (test-engineer)

• Create E2E tests for happy path

Infrastructure Tasks (infrastructure-engineer)

• [List infrastructure tasks if any, or mark N/A]

Review Tasks (code-reviewer)

• Review code quality and standards
• Ensure 80-90% test coverage
• Check security implementation

Base this on the specification in docs/tickets/$ARGUMENT/specification.md"

*Mark "Generate technical specification and tasks" as completed*

## PHASE 3: Parallel Implementation
*Mark "Parallel implementation (engineering, testing, infrastructure)" as in_progress*

### Step 3.1: Launch Parallel Implementation [EXECUTE IN PARALLEL]

LAUNCH 3 AGENTS IN PARALLEL:

AGENT 1: full-stack-engineer PROMPT FOR AGENT: "Implement all engineering tasks for issue #$ARGUMENT:

1. Read docs/tickets/$ARGUMENT/specification.md
2. Read docs/tickets/$ARGUMENT/tasks.md
3. Implement ALL full-stack-engineer tasks
4. AS YOU COMPLETE EACH TASK:
   • Update docs/tickets/$ARGUMENT/tasks.md
   • Change '- [ ]' to '- [x]'
5. Write unit tests (co-located .test.ts files)
6. Ensure >80% test coverage
7. VERIFY all your tasks are marked [x]"

AGENT 2: test-engineer PROMPT FOR AGENT: "Implement E2E tests for issue #$ARGUMENT:

1. Read docs/tickets/$ARGUMENT/tasks.md
2. Create E2E tests for happy path in e2e-tests/
3. Include accessibility tests using axe-core
4. Update docs/tickets/$ARGUMENT/tasks.md marking tests as [x]
5. DO NOT run tests yet - will run in next phase"

AGENT 3: infrastructure-engineer PROMPT FOR AGENT: "Handle infrastructure for issue #$ARGUMENT:

1. Read docs/tickets/$ARGUMENT/specification.md
2. If infrastructure section exists, implement changes
3. If no infrastructure needed, mark tasks as [x] or N/A
4. Update docs/tickets/$ARGUMENT/tasks.md
5. DO NOT run migrations - will run in next phase"

WAIT FOR ALL AGENTS TO COMPLETE

*Mark "Parallel implementation (engineering, testing, infrastructure)" as completed*

## PHASE 4: Testing and Quality
*Mark "Run all tests and quality checks" as in_progress*

### Step 4.1: Run Test Suite

EXECUTE IN SEQUENCE:

1. yarn lint
2. yarn dev (verify app boots - kill after 10 seconds if successful)
3. yarn test (unit tests)
4. yarn test:e2e (E2E tests)

IF ANY FAILURES:

• Use test-engineer to fix e2e tests, use full-stack-engineer agent to fix everything else
• Re-run failed tests
• Repeat until all pass

VERIFY: All tests passing

*Mark "Run all tests and quality checks" as completed*

## PHASE 5: Code Review
*Mark "Code review and final verification" as in_progress*

### Step 5.1: Automated Review

AGENT: code-reviewer TASK: Comprehensive code review PROMPT FOR AGENT: "Review implementation of issue #$ARGUMENT:

1. Run git diff master to see all changes
2. Check CLAUDE.md adherence:
   • Naming conventions
   • Module structure
   • Welsh translations included
   • No business logic in apps/