Skip to content

GitHub Copilot

VS Code GitHub Copilot configuration with custom agents. Config managed with stow.

Installation

cd ~/dotfiles
stow copilot

Symlinks ~/.copilot/ to dotfiles.

Structure

copilot/.copilot/
├── agents/            # Custom agent definitions
├── skills/            # Reusable skills
└── mcp.json           # MCP server configuration

Agents

Primary Agents

Agent Model Use Case
turbo default Fast iteration, autonomous execution
ideation default Brainstorming, creative exploration
airflow-migration-orchestrator GPT-5.2 Large-scale DAG migrations

Subagents

Agent Model Use Case
builder GPT-5.2-Codex Complex multi-file implementations
quick-fix Claude Haiku 4.5 Small one-off edits (docstrings, renames, imports)
tester GPT-5.2-Codex Test writing and validation
airflow-expert GPT-5.2-Codex DAG development and debugging
code-summarizer Claude Haiku 4.5 Code and project summarization
ruthless-reviewer default Code review, security audit
planner default Task decomposition and planning

Model Strategy

  • Quality tasks (builder, tester, airflow-expert): GPT-5.2-Codex - latest code-optimized model
  • Speed tasks (quick-fix, code-summarizer): Claude Haiku 4.5 - fast, efficient for focused work
  • Reasoning tasks (ruthless-reviewer, planner): default (Sonnet) - deeper analysis needed

Agent Frontmatter

Copilot agents use simplified frontmatter compared to OpenCode:

---
name: "builder"
description: "Python/K8s/AWS expert that builds with minimal code"
model: github-copilot/gpt-5.2-codex
tools: ['read', 'edit', 'execute', 'search', 'vscode', 'terminal']
---

Key Differences from OpenCode

Feature OpenCode Copilot
Agent identifier description name
Mode setting mode: primary/subagent Not supported
Permissions Detailed permission: block Not supported
Temperature temperature: 0.1 Not supported
Tools OpenCode-specific names VS Code-specific names

Skills

Reusable knowledge modules:

Skill Purpose
python-uv PEP 723 scripts, uv-first policy
airflow DAG patterns, KubernetesPodOperator
k8s-troubleshoot Kubernetes debugging
aws-iam IAM policy creation and debugging
steampipe-aws AWS resource discovery via SQL
code-review Ruthless review methodology
minimal-build Build with minimal code

Note: progress-tracker skill exists in OpenCode but not Copilot (requires todowrite tool).

MCP Servers

Configured in mcp.json:

{
  "mcp-server-docker": {
    "type": "local",
    "command": "uvx",
    "args": ["mcp-server-docker"],
    "disabled": true
  }
}
Server Purpose
mcp-server-docker Docker operations
awesome-copilot Additional Copilot tools

Usage

Invoke agents in Copilot Chat:

@builder implement the user authentication module
@quick-fix add a docstring to the process_data function
@ruthless-reviewer check this PR for security issues
@planner break down this feature into subtasks

Syncing with OpenCode

The agents are kept in sync between OpenCode and Copilot. Key differences:

  1. Naming: Some agents have different names
  2. OpenCode summarizer → Copilot code-summarizer
  3. OpenCode reviewer → Copilot ruthless-reviewer

  4. Frontmatter: Copilot uses simpler schema (no permissions, temperature)

  5. Tools: Tool names differ between platforms

  6. Skills: progress-tracker is OpenCode-only

Tips

  1. Use @quick-fix for small edits, @builder for complex work
  2. Check justfile first for project commands
  3. Use @planner to break down complex tasks before implementation
  4. The planner knows about all available agents and when to use each

See Also

  • OpenCode - Terminal-based configuration (source of truth)
  • VS Code - General VS Code setup