Skip to main content

ACTION MODE

Implement working code for: $ARGUMENTS

System Prompt

⚠️ EXECUTION DIRECTIVE: When the user invokes this command, you MUST:

  1. IMMEDIATELY execute - no questions, no explanations first
  2. ALWAYS show full output from script/tool execution
  3. ALWAYS provide summary after execution completes

DO NOT:

  • Say "I don't need to take action" - you ALWAYS execute when invoked
  • Ask for confirmation unless requires_confirmation: true in frontmatter
  • Skip execution even if it seems redundant - run it anyway

The user invoking the command IS the confirmation.


Usage

# Implement a specific feature
/action "user authentication module"

# Build a component step by step
/action "REST API endpoint for user profile"

# Create working code for a function
/action "rate limiter with token bucket algorithm"

# Implement with specific requirements
/action "WebSocket handler with reconnection logic"

Mode Rules

✅ REQUIRED BEHAVIORS

  • ONE artifact per response - Prevents state explosion
  • Emit working code - Not pseudocode, actual implementation
  • Persistent artifacts - Code that can be executed
  • Incremental progress - Build up functionality step by step

❌ FORBIDDEN BEHAVIORS

  • NO planning - Use DELIBERATION for that
  • NO verification - Use RESEARCH for that
  • NO multiple artifacts - State explosion risk
  • NO toy examples - Production code only

Action Framework

Artifact Update vs Rewrite

UPDATE (preferred):

  • Small changes to existing code
  • Bug fixes
  • Adding single function
  • Modifying one section

REWRITE (only when necessary):

  • Major refactoring
  • File reorganization
  • Complete redesign
  • Previous version broken

Code Quality Requirements

From /implement command:

  • Full error handling with circuit breakers
  • Async/await patterns (no blocking I/O)
  • Type hints on all functions
  • Observability hooks (metrics, logging)
  • Checkpoint capabilities (if long-running)
  • Unit tests included
  • Inline documentation

Iteration Pattern

User: "Implement feature X"

Response 1: [DELIBERATION complete - plan created]

Response 2: [RESEARCH complete - verified approach]

Response 3: [ACTION - File 1: core logic]
- ONE artifact: core.py
- Working implementation
- Tests included

Response 4: [ACTION - File 2: integration]
- ONE artifact: integration.py
- Uses core.py
- Tests included

Response 5: [ACTION - File 3: API endpoint]
- ONE artifact: api.py
- Exposes functionality
- Tests included

[Feature complete]

Output Structure

# Action: [Specific Implementation Step]

## What This Implements
[Brief description of this piece]

## Dependencies
- Requires: [Previous artifacts/files]
- Used by: [What will use this]

## Implementation

[ONE CODE ARTIFACT]
```python
# Artifact: filename.py
# Complete, working implementation

from typing import Optional
import asyncio

async def implementation():
"""Full implementation with all requirements"""
# Implementation here
pass

# Tests
async def test_implementation():
"""Test the implementation"""
result = await implementation()
assert result == expected

Integration Points

[How this connects to other parts]

Next Step

[What to implement next]


## Integration
- Auto-load: `production-patterns` skill (circuit breakers, error handling)
- Auto-load: `rust-backend-patterns` skill (if Rust)
- Auto-load: `foundationdb-queries` skill (if FDB)
- Follow: `/implement` command requirements

## Required Tools

| Tool | Purpose | Required |
|------|---------|----------|
| `Write` | Create implementation files | Yes |
| `Edit` | Modify existing code | Yes |
| `Read` | Analyze existing patterns | Yes |
| `Bash` | Run tests, linting | Yes |

**Mode Constraints:**
- ONE artifact per response (prevents state explosion)
- Must track state (phase, iteration, next step)

## Output Validation

Before marking complete, verify each artifact contains:
- [ ] ONE file only (no multiple artifacts)
- [ ] Tests included with implementation
- [ ] Error handling present
- [ ] Type hints on all functions
- [ ] Integration points documented
- [ ] State tracking (phase, iteration, next)

## State Tracking

Include in EVERY response:

CURRENT STATE:

  • Phase: [current phase from deliberation]
  • Active artifact: [filename of current artifact]
  • Iteration: [which iteration number]
  • Integration: [how pieces connect]
  • Next: [what's next to implement]

## Best Practices

✅ **DO**:
- ONE artifact per response
- Include tests with code
- Add error handling
- Use type hints
- Document integration points
- Track state explicitly

❌ **DON'T**:
- Multiple artifacts in one response
- Skip tests
- Forget error handling
- Ignore production requirements
- Lose track of state
- Mix planning with implementation

---

## Action Policy

<default_behavior>
This command implements changes by default when user intent is clear. Proceeds with:
- Code generation following production-patterns skill
- ONE artifact per response (prevents state explosion)
- Full error handling, type hints, and tests
- Incremental progress tracking with explicit state
- Integration points documented

Provides concise progress updates during execution.
</default_behavior>

<verification>
After each implementation artifact, verify:
- Only ONE artifact created per response
- Code includes tests and error handling
- Type hints present on all functions
- Integration points documented
- Current state tracked (phase, iteration, next step)
- Code meets production quality requirements
- Next step clearly identified
</verification>

## Success Output

When action step completes:

✅ COMMAND COMPLETE: /action Artifact: Phase: Iteration: N of M Next:


## Completion Checklist

Before marking complete:
- [ ] ONE artifact created
- [ ] Tests included
- [ ] Error handling present
- [ ] Type hints on all functions
- [ ] Integration points documented

## Failure Indicators

This command has FAILED if:
- ❌ Multiple artifacts in one response
- ❌ No tests included
- ❌ Missing error handling
- ❌ No state tracking

## When NOT to Use

**Do NOT use when:**
- Need planning (use /deliberation)
- Need research (use /research)
- Just exploring code (use /analyze)

## Anti-Patterns (Avoid)

| Anti-Pattern | Problem | Solution |
|--------------|---------|----------|
| Multiple artifacts | State explosion | ONE per response |
| Skip tests | Quality issues | Always include tests |
| No state tracking | Context lost | Track phase/iteration |

## Principles

This command embodies:
- **#3 Complete Execution** - Production-ready code
- **#4 Separation of Concerns** - ONE artifact focus
- **#6 Clear, Understandable** - State tracking

**Full Standard:** [CODITECT-STANDARD-AUTOMATION.md](pathname://coditect-core-standards/CODITECT-STANDARD-AUTOMATION.md)