copilot-cli

name	copilot-cli
description	Executes tasks using the GitHub Copilot CLI (`copilot`). Automatically determines the safest permission mode (plan, editor, or autopilot) based on the task type.

executing-copilot

Purpose

Use this skill to perform coding, research, or automation tasks using the copilot CLI. It maps requests to Copilot's native tool-permission flags to balance automation with security.

Permission Tiers

Tier	Mode	Flag(s)	Capability	Approval Required	Typical Tasks
0	plan	(none)	Read-only analysis; prompts for every tool.	No	Code review, audits, research.
1	editor	--allow-tool='write'	Auto-approves file edits; prompts for shell.	Yes	Refactoring, documentation, lint fixes.
2	autopilot	--allow-all-tools	Auto-approves ALL tools (edits + shell).	Yes (High Risk)	CI/CD, complex builds, automated testing.

Implementation Workflow

1. Analyze & Classify

Analyze the user's intent to determine the required permission tier.

• Tier 0 (Plan): Does the task only involve reading code, analyzing architecture, or gathering information?
• Tier 1 (Editor): Does the task involve modifying files but no autonomous shell command execution?
• Tier 2 (Autopilot): Does the task require running tests, build scripts, or managing dependencies without interruption?

2. Approval Protocol

If the task maps to Tier 1 or Tier 2, you MUST obtain user approval before executing the copilot command.

Use the AskQuestion tool to confirm:

"I've detected that this task requires [Editor/Autopilot] permissions to [modify files/run shell commands]. OK to proceed?"

3. Execution

Execute copilot using the flags determined by the tier. Use -p for non-interactive (programmatic) execution suitable for the calling agent.

# Tier 0 (Plan — read-only analysis)
copilot -p "<prompt>"

# Tier 1 (Editor — file edits auto-approved, shell still prompts)
copilot -p "<prompt>" --allow-tool='write'

# Tier 2 (Autopilot — all tools auto-approved)
copilot -p "<prompt>" --allow-all-tools

Security Rules:

• NEVER use --allow-all-tools without explicit user confirmation of the risks.
• ALWAYS use the most restrictive permission level possible (prefer no flags / Tier 0).
• To block a specific dangerous tool, add --deny-tool='<tool>' to any command.
• If you are unsure, default to Tier 0 and escalate only if copilot reports it cannot complete the task.

4. Advanced Tool Control

For fine-grained control, combine --allow-tool and --deny-tool flags:

# Allow shell commands but deny network access
copilot -p "<prompt>" --allow-tool='shell' --deny-tool='url'

# Allow a specific MCP server tool only
copilot -p "<prompt>" --allow-tool='mcp-server-name'

Configuration

This skill leverages native copilot CLI flags to enforce permission tiers. Global defaults can be set in ~/.copilot/config.json. Authentication requires running copilot auth login or having a valid GITHUB_TOKEN in the environment.

Examples

Refer to references/usage-examples.md for concrete scenarios.