SkillAgentSearch skills...

implement

Execute phased implementation with validation gates

Install / Use

npx skills add shanraisshan/claude-code-best-practice

Installs into whichever agent you are using.

About this skill
⚡

Claude Commands

Claude Code slash commands

Quality Score

93/100

Supported Platforms

Claude Code
Zed

Our assessment of implement

implement scores 93/100 on our quality scale, 66th of 398 Content & Media skills we index (top 17%).

Its Claude Commands is 16 KB long, well organised into 77 sections with 13 code examples: a thorough specification that gives an agent plenty to work with.

With 66,294 GitHub stars, it is one of the more widely adopted skills in the catalogue.

Substance
30/30
Structure
20/20
Description
8/15
Adoption
20/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated 2 days ago, so implement is actively maintained.
  • It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
  • Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

implement compared with similar skills

All 4 of these similar skills score higher than implement; compare them before choosing.

SkillScoreStarsUpdatedFormat
implement (this skill)by shanraisshan9366.3k2d agoClaude Commands
claude-memby thedotmack10094.7ktodayCLAUDE.md
Agent-Reachby Panniantong10085.5k10d agoCLAUDE.md
Understand-Anythingby Egonex-AI10084.2k14d agoCLAUDE.md
headroomby headroomlabs-ai10073.8ktodayCLAUDE.md

Frequently asked questions

How do I install implement?
Run npx skills add shanraisshan/claude-code-best-practice. The install tabs above show the steps for each supported agent.
Which AI agents does implement work with?
It is written for Claude Code and Zed, as a Claude Commands file. Other agents that read the same format can often use it too.
Is implement safe to use?
It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is implement still maintained?
The repository was last updated 2 days ago, so implement is actively maintained.

description: Execute phased implementation with validation gates argument-hint: "<feature-slug> [--phase N] [--validate-only]"

User Input

$ARGUMENTS

You MUST parse the user input to extract the feature slug (the folder name in rpi/).

Purpose

This command executes phased implementation of features based on planning documentation. It orchestrates specialized agents, enforces validation gates, and ensures constitutional compliance throughout implementation.

Prerequisites:

  • Feature folder exists at rpi/{feature-slug}/
  • Planning completed (rpi/{feature-slug}/plan/PLAN.md exists)

Output Location: rpi/{feature-slug}/implement/

This is Step 4 of the RPI Workflow (final step - actual implementation).

Flags

  • --phase N: Execute specific phase number (1-8), if omitted starts from phase 1
  • --validate-only: Only validate current phase, don't implement
  • --skip-validation: Skip validation gate and proceed (use with caution)

Available Agents

All agents use Opus model for maximum quality.

Implementation Agent

| Agent | Type | When to Use | |-------|------|-------------| | senior-software-engineer | Custom | All implementation tasks |

Support Agents

| Agent | Type | Purpose | |-------|------|---------| | Explore | Built-in | Pre-implementation code exploration | | code-reviewer | Custom | Code review and quality validation | | constitutional-validator | Custom | Validate against project constitution | | documentation-analyst-writer | Built-in | Documentation generation |

Agent Routing

All implementation tasks are handled by the senior-software-engineer agent.


Phase 0: Load Context and Rules

Prerequisites: Feature slug parsed from user input

Process:

0.1 Load Project Constitution

  1. Check for a constitution or principles document in the repository
  2. If exists, extract:
    • Technical constraints (type safety, testing, component isolation)
    • Business principles (quality standards, workflow)
    • Architectural boundaries
  3. Store constraints for enforcement during implementation

0.2 Load Domain-Specific Guidelines

Based on files to be modified, load relevant project guidelines:

  • Check for component-specific README files
  • Check for coding style guides
  • Check for testing requirements documentation

0.3 Analyze Implementation Scope

  1. Read rpi/{feature-slug}/plan/PLAN.md
  2. Identify all files to be modified
  3. Map files to implementation agent

Outputs:

  • Constitutional context summary
  • Domain rules loaded
  • File-to-agent mapping
  • Phase execution plan

Validation:

  • [ ] Constitution loaded (if exists)
  • [ ] Domain rules loaded for affected files
  • [ ] All files mapped to agents
  • [ ] Execution plan understood

Phased Implementation Workflow

Phase Implementation Loop

For each phase in PLAN.md:

┌─────────────────────────────────────────────────────────────────┐
│ Phase N: [Phase Name]                                            │
├─────────────────────────────────────────────────────────────────┤
│                                                                  │
│  1. Code Discovery (Explore Agent)                              │
│     └─→ Understand existing code before changing it             │
│                                                                  │
│  2. Implementation (senior-software-engineer)                   │
│     └─→ Implement phase deliverables                            │
│                                                                  │
│  3. Self-Validation                                             │
│     └─→ Engineer validates against phase checklist              │
│                                                                  │
│  4. Code Review (code-reviewer Agent)                           │
│     └─→ Security, correctness, maintainability                  │
│                                                                  │
│  5. User Validation Gate                                        │
│     └─→ STOP and request user approval                          │
│         ├─→ PASS: Proceed to next phase                         │
│         ├─→ CONDITIONAL PASS: Note issues, proceed              │
│         └─→ FAIL: Fix issues, re-validate                       │
│                                                                  │
│  6. Documentation Update                                        │
│     └─→ Update phase status in PLAN.md                          │
│                                                                  │
└─────────────────────────────────────────────────────────────────┘

Step 1: Code Discovery (Per Phase)

Agent: Explore (Built-in, via Task tool)

Purpose: Ground implementation in code reality before making changes.

Process:

  1. Launch Explore agent via Task tool with subagent_type="Explore"
  2. Request analysis of files affected by current phase
  3. Understand existing patterns, integration points, constraints

Explore Agent Prompt:

Analyze the codebase to prepare for implementing Phase N of [feature-name].

Files to be modified in this phase:
[List files from PLAN.md]

Investigate and document:

1. **Current Implementation**
   - How do these files currently work?
   - What patterns are used?
   - What functions/classes will be affected?

2. **Integration Points**
   - What other files import or use these modules?
   - What APIs or interfaces will change?
   - What tests cover this code?

3. **Dependencies**
   - What libraries are used?
   - What internal utilities are available?
   - What constraints exist from current code?

4. **Patterns to Follow**
   - What coding style is used in these files?
   - What naming conventions are followed?
   - What error handling patterns exist?

5. **Risks and Considerations**
   - What could break if we change this?
   - What edge cases exist?
   - What backward compatibility concerns?

Provide a discovery summary to inform implementation.

Output: Discovery summary for implementation agent


Step 2: Implementation (Per Phase)

Agent: senior-software-engineer

Process:

  1. Use senior-software-engineer agent
  2. Provide discovery context from Step 1
  3. Implement all deliverables for the phase
  4. Follow constitutional constraints and project rules

Implementation Agent Prompt Template:

Acting as the [agent-name] agent, implement Phase N deliverables for [feature-name].

## Context
- Constitutional Constraints: [from Phase 0]
- Domain Rules: [from Phase 0]
- Discovery Summary: [from Step 1]

## Phase N Deliverables
[List from PLAN.md]

## Files to Modify
[List files with specific changes from PLAN.md]

## Implementation Requirements
1. Follow existing code patterns identified in discovery
2. Honor constitutional constraints (type safety, testing, etc.)
3. Follow project-specific rules (if applicable)
4. Write tests for new functionality
5. Include appropriate logging
6. Handle errors gracefully

## Quality Checklist
- [ ] Code follows existing patterns
- [ ] Type annotations present where applicable
- [ ] Tests written and passing
- [ ] No breaking changes to existing functionality
- [ ] Logging added for observability
- [ ] Error handling comprehensive

Implement all deliverables and report what was done.

Step 3: Self-Validation

Agent: senior-software-engineer (same as Step 2)

Process:

  1. Agent validates implementation against phase checklist
  2. Run linting (use project's configured linter)
  3. Run tests relevant to changes
  4. Verify build succeeds

Validation Commands (adjust to your project):

# Run linter
[your-linter-command]

# Run tests
[your-test-command]

# Build/compile
[your-build-command]

Self-Validation Checklist:

  • [ ] All deliverables implemented
  • [ ] Linting passes
  • [ ] Tests pass
  • [ ] Build succeeds
  • [ ] No regressions in existing tests
  • [ ] Constitutional constraints honored
  • [ ] Domain rules followed

Step 4: Code Review

Agent: code-reviewer (Custom, auto-invoked)

Process:

  1. Invoke code-reviewer agent to review changes
  2. Focus on correctness, security, maintainability
  3. Address blockers before proceeding

Code Review Agent Prompt:

Acting as the code-reviewer agent, review the Phase N implementation for [feature-name].

## Files Changed
[List modified files]

## Changes Made
[Summary of implementation]

## Review Focus
- Correctness & tests
- Security & dependency hygiene
- Architectural boundaries
- Clarity over cleverness

## Constitutional Constraints
[From Phase 0]

Provide review using standard output format.

Review Verdicts:

  • APPROVED: Proceed to user validation
  • APPROVED WITH SUGGESTIONS: Note suggestions, proceed
  • NEEDS REVISION: Fix issues, re-review

Step 5: User Validation Gate

CRITICAL: This step REQUIRES user interaction. DO NOT proceed automatically.

Process:

  1. Present phase deliverables checklist
  2. Show what was implemented (files changed, features added)
  3. Present validation criteria from PLAN.md
  4. Show code review results
  5. STOP and wait for user decision

Validation Request Format:

## Phase N Validation Request

### Deliverables Completed
- [x] [Deliverable 1] - [implementation summary]
- [x] [Deliverable 2] - [implementation summary]
- ...

### Files Changed
| File | Change Type | Lines |
|------|-------------|-------|
| [file] | [add/modify] | [±N] |

### Tests
- [x] Unit tests: PASS
- [x] Integration tests: PASS
- [x] Build: SUCCESS

### Code Review
- Verdict: [APPROVED / APPROVED WITH SUGGESTIONS]
- Issues: [None / List]

### Validation Criteria (from PLAN.md)
- [ ] [Criterion 1]
- [ ] [Criterion 2]
- ...

---

**Please validate Phase N:**
- **PASS**: Phase complete, proceed to Phase N+1
- **CONDITIONAL PASS**: Note issues below, proceed with caution
- **FAIL**: Specify issues to fix before proceeding

User Decisions:

  • PASS: Proceed to next phase
  • CONDITIONAL PASS: Document issues, proceed to next phase
  • FAIL: Fix issues, re-run Steps 2-5

Step 6: Documentation Update

Process:

  1. Update rpi/{feature-slug}/plan/PLAN.md with phase status
  2. Update rpi/{feature-slug}/implement/IMPLEMENT.md with validation results
  3. Append each phase's validation to IMPLEMENT.md

Phase Status Tracking

Update checkboxes in PLAN.md:

- [ ] Phase N: Not Started
- [~] Phase N: In Progress
- [x] Phase N: Validated (PASS)
- [!] Phase N: Conditional Pass (with notes)
- [-] Phase N: Failed Validation (needs rework)

IMPLEMENT.md Template

# Implementation Record

**Feature**: [feature-slug]
**Started**: [Date]
**Status**: [IN_PROGRESS / COMPLETED]

---

## Phase 1: [Phase Name]

**Date**: [Date]
**Verdict**: [PASS / CONDITIONAL PASS / FAIL]

### Deliverables
- [x] [Deliverable 1]
- [x] [Deliverable 2]

### Files Changed
[List with line counts]

### Test Results
[Test output summary]

### Code Review
[Review verdict and notes]

### Notes
[Any additional notes]

---

## Phase 2: [Phase Name]
[Same structure as Phase 1...]

---

## Summary

**Phases Completed**: [N] of [N]
**Final Status**: [COMPLETED / IN_PROGRESS]

Error Handling

Implementation Failures

If implementation fails:

  1. Document the specific failure
  2. Analyze root cause
  3. Try alternative approach (max 2 attempts)
  4. If still failing, STOP and ask user for guidance
  5. Do NOT proceed to next phase with broken implementation

Message: "Implementation failed: [error]. Attempted [N] approaches. User guidance needed."

Test Failures

If tests fail:

  1. Analyze failure cause (code bug vs test bug)
  2. Fix the issue
  3. Re-run tests
  4. If persistent, document and ask user
  5. Do NOT mark phase complete with failing tests

Message: "Tests failing: [failures]. Fix attempted but unsuccessful. User review needed."

Build Failures

**If build

Truncated for display — read the full file on GitHub.

Related Skills

View on GitHub
GitHub Stars66.3k
CategoryContent
Updated2d ago
Forks6.6k

Languages

HTML

Trust signals

100/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

No cautions