// Scaffold a new custom agent — reviewers, researchers, or workflow agents. Use when asked to create, scaffold, or add a new agent.
Explore requirements and approaches through collaborative dialogue before writing a right-sized requirements document and planning implementation. Use for feature ideas, problem framing, when the user says 'let's brainstorm', or when they want to think through options before deciding what to build.
Document a solved problem as reusable institutional knowledge with parallel research. Use when the user says 'document this', 'compound', 'save learnings', or wants to capture a solution for future reference.
Enhance an existing plan with parallel research agents for depth, best practices, and implementation details. Use when the user says 'deepen the plan', 'research more', or when a plan has high-risk dimensions that need more investigation.
Review requirements or plan documents using parallel persona agents that surface role-specific issues. Use when a requirements document or plan document exists and the user wants to improve it, or when invoked by /cplan for quality gating.
Commit, push, and open a PR with an adaptive, value-first description. Use when the user says "commit and PR", "push and open a PR", "ship this", "create a PR", "open a pull request", or wants to go from working changes to an open pull request in one step. Also handles "update the PR description" or "refresh the PR description".
Create a git commit with a clear, value-communicating message. Use when the user says "commit", "commit this", "save my changes", "create a commit", or wants to commit staged or unstaged work. Produces well-structured commit messages that follow repo conventions when they exist, and defaults to conventional commit format otherwise.
| name | ccreate-agent |
| description | Scaffold a new custom agent — reviewers, researchers, or workflow agents. Use when asked to create, scaffold, or add a new agent. |
| argument-hint | Describe the agent to create (e.g., 'Vue reviewer', 'Python linter') |
Scaffold a new .agent.md file with correct VS Code Copilot frontmatter and structure.
Ask the user what kind of agent they want to create:
creview orchestrator)Based on agent type, ask:
vue-reviewer, python-linter)Reviewers are subagents dispatched by the creview orchestrator. They:
---
name: [name]-reviewer
description: "[What it reviews] — [specific focus areas]"
user-invocable: false
disable-model-invocation: true
tools: ['search/codebase', 'search/usages']
---
The body should include:
Researchers are read-only subagents used by other agents for codebase exploration.
---
name: [name]
description: "[What it researches]"
user-invocable: false
disable-model-invocation: true
tools: ['search/codebase', 'search/usages', 'search/changes', 'web/fetch']
---
Workflow agents are user-facing and can have full tool access.
---
name: [name]
description: "[What it does]"
argument-hint: "[Expected input]"
tools: ['*']
---
Write the agent file to .github/agents/[name].agent.md.
For reviewers, explain:
creview orchestrator uses agents: ['*'] and auto-discovers all *-reviewer.agent.md filesdescription to determine if it's relevant to the current diff| Field | Type | Notes |
|---|---|---|
name | string | Agent identifier (lowercase, hyphenated) |
description | string | Shown in chat input; used for smart routing |
argument-hint | string | Guidance text for input |
tools | string[] | Tool/tool-set names |
agents | string[] or * | Subagent availability |
user-invocable | boolean | Show in dropdown (default: true) |
disable-model-invocation | boolean | Prevent auto-invocation (default: false) |
handoffs | object[] | Workflow transitions |
Banned fields (Claude Code only — must not appear):
color, context, allowed-tools, disallowedTools
Deprecated fields:
infer — use user-invocable and disable-model-invocation instead