# Coordinate Claude and Copilot Delivery

Complex coding tasks often lose structure across planning, implementation, and review. This skill defines a staged Claude and Copilot workflow with clear artifacts.

## Install

```bash
npx skillstore add aykahshi/copilot-flow
```

## Metadata

- - Slug: aykahshi-copilot-flow
- - Version: 1.0.0
- - Author: Aykahshi
- - GitHub username: Aykahshi
- - License: MIT
- - Repository: https://github.com/Aykahshi/copilot-mcp-tool/tree/main/plugins/copilot-flow/skills/copilot-flow-integration
- - Ref: main
- - Supported tools: Claude, Codex, Claude Code
- - Risk level: medium
- - Risk factors: network, filesystem
- - Quality score: 50
- - Quality tier: warning
- - Public page: https://skillstore.pages.dev/skills/aykahshi-copilot-flow
- - Manifest: https://skillstore.pages.dev/api/skills/aykahshi-copilot-flow/manifest

## Capabilities

- Defines a five-stage workflow: analyze, design, implement, review, and deliver.
- Documents slash commands for running the full flow or individual stages.
- Assigns analysis, implementation, and delivery work to Claude.
- Assigns architecture design and code review work to Copilot through MCP.
- Specifies expected artifacts such as analysis reports, design notes, review reports, and delivery packages.
- Describes workflow state recovery through a local state file.

## Use Cases

- Plan a Feature Build: Break a feature request into analysis, architecture, implementation, review, and final delivery steps.
- Add Independent Review: Use Copilot review stages to check code quality, security, performance, and best practices after Claude implements changes.
- Recover Interrupted Work: Resume a multi-stage task from documented workflow state and previously generated stage artifacts.

## Prompt Templates

### Start a Full Workflow

```
Run copilot-flow for this task: build a small user profile editing feature. Show the planned stages before making changes.
```

### Analyze Requirements Only

```
Use the analyze stage for this request: improve our password reset flow. Produce clear requirements and open questions.
```

### Request Architecture Design

```
Use the design stage with Copilot MCP. Create architecture guidance for the analyzed notification service changes.
```

### Run Review and Delivery

```
Run the review stage for the recent API changes, then prepare delivery notes with risks, tests, and follow-up work.
```

## Limitations

- The skill file is documentation and does not include the command implementations.
- Copilot MCP and claude-mem must be configured separately.
- The workflow can modify project files when implementation stages run.
- Users must review outputs for correctness, security, and project fit.

## Best Practices

- Start with clear requirements and constraints before running the full workflow.
- Review each generated artifact before allowing the next stage to change files.
- Use individual stages for focused analysis, design, implementation, or review tasks.

## Anti Patterns

- Do not run the implementation stage on repositories with uncommitted critical work.
- Do not send confidential code or data to Copilot MCP without policy approval.
- Do not skip review when the workflow changes authentication, security, or data handling code.

## Security Audit

- - Safe to publish: true
- - Audited at: 2026-06-28T12:49:29.951\+00:00
- - Summary: Static analysis reported many high-risk patterns, but the reviewed file contains Markdown documentation rather than executable code. The backtick, weak cryptography, and reconnaissance detections are false positives; the real concerns are repository file modification, external Copilot MCP collaboration, and possible interaction history retention.

## Stats

- - Views: 229
- - Downloads: 5
- - Favorites: 0
- - Popularity score: 0