AGENT SKILL

agent-management

Creates, updates, and deprecates agent definitions for both Claude Code (.claude/agents/*.md) and GitHub Copilot/VS Code (.github/agents/*.agent.md). Invoked exclusively by the agent-manager agent.

License 许可
GPL-3.0
最近更新
2026-06-06
Stars
★ 0
Forks
0

01 前言 Front Matter

name
agent-management
description
Creates, updates, and deprecates agent definitions for both Claude Code (.claude/agents/*.md) and GitHub Copilot/VS Code (.github/agents/*.agent.md). Invoked exclusively by the agent-manager agent.

03 SKILL.md 正文

Agent Management

You scaffold and maintain all agent definitions across both platforms. Parse the args to determine the operation mode, then execute the corresponding procedure.


Phase 0 — Context Load (silent)

  1. Read .claude/CLAUDE.md to internalize project conventions and portability rules.
  2. Invoke Skill("manage-memory", args: "agent-manager") to load persistent memory.
  3. Glob .claude/agents/*.md and .github/agents/*.agent.md to build the current agent roster.

Mode: list

Args: list

  1. Glob .claude/agents/*.md — collect name, model, tools, and description from frontmatter.
  2. Glob .github/agents/*.agent.md — collect name from frontmatter.
  3. Cross-reference: identify agents that exist in one platform but not the other (orphans).
  4. Return a formatted table with columns: agent, claude-code, copilot, model, status.

Mode: create-agent

Args: create-agent <name>

  1. Validate <name> is kebab-case.
  2. Fail if .claude/agents/<name>.md already exists.
  3. Collect required inputs:
    • model — default claude-sonnet-4-6
    • color — default cyan
    • tools — default Read, Glob, Grep, Skill, TodoWrite
    • One-line purpose describing the agent's domain
    • 2–4 bullet responsibilities
  4. Write .claude/agents/<name>.md using the Claude Agent Template below.
  5. Write .github/agents/<name>.agent.md using the Copilot Agent Template below — apply the Claude-only Tool Exclusion List (see Mode: sync-agent) when populating its tools: frontmatter.
  6. Create .claude/agent-memory/<name>/ directory: a. Windows: mkdir -p ".claude/agent-memory/<name>" b. Linux/Mac: mkdir -p ".claude/agent-memory/<name>/"
  7. Write .claude/agent-memory/<name>/MEMORY.md using the Memory Index Template below.
  8. Confirm: "Agent '<name>' created in .claude/agents/ and .github/agents/."

Claude Agent Template

---
name: "<name>"
description: "<multi-line description with at least one <example> block>"
tools: Read, Glob, Grep, Skill, TodoWrite
model: <model>
color: <color>
memory: project
---

You are the **<Title>** agent for the HypermediaEngine project.

## Responsibilities

- [Responsibility 1]
- [Responsibility 2]
- [Responsibility 3]

## Skills

### `manage-memory` — invoke at session start and when learning something worth preserving

Skill("manage-memory", args: "<name>") // load Skill("manage-memory", args: "save <name> ...") // save


Record: [what this agent should remember across sessions]

### `agent-management` — route all agent definition changes through agent-manager

To create or update an agent definition:

Agent("agent-manager", prompt: "create-agent <name>") Agent("agent-manager", prompt: "update-agent <name> <change-description>")


## Protocols

- [Protocol 1]
- [Protocol 2]

Copilot Agent Template

---
name: "<name>"
description: "<one-line description for Copilot agent picker>"
tools: [read, edit, search, todo]
user-invocable: true
---
You are the <Title> for the HypermediaEngine project.

## Responsibilities
1. [Responsibility 1]
2. [Responsibility 2]

## Preferred Skills
- `<primary-skill>`
- `manage-memory`

Memory Index Template

# Memory Index


Mode: update-agent

Args: update-agent <name> <change-description>

  1. Validate that both .claude/agents/<name>.md and .github/agents/<name>.agent.md exist; warn on missing platform file.
  2. Parse <change-description> to determine what to change (frontmatter field, responsibilities, skills section, protocols, etc.).
  3. Apply changes to .claude/agents/<name>.md using Edit.
  4. Apply equivalent changes to .github/agents/<name>.agent.md using Edit.
  5. If the change touches the tools: frontmatter, apply the Claude-only Tool Exclusion List (see Mode: sync-agent) before writing the Copilot file — never propagate excluded tools.
  6. Validate Claude agent frontmatter still contains: name, description, tools, model.
  7. Validate Copilot agent frontmatter still contains: name, description, tools.
  8. Validate Copilot tools: array contains none of the excluded Claude-only tool names.
  9. Confirm: "Agent '<name>' updated on both platforms."

Mode: sync-agent

Args: sync-agent <name>

Use when a Claude agent definition exists but the Copilot counterpart is missing or out of date.

  1. Read .claude/agents/<name>.md.
  2. Derive the equivalent .github/agents/<name>.agent.md content from it using the Copilot Agent Template.
  3. Apply the Claude-only Tool Exclusion List (see below) when translating the tools: frontmatter — strip excluded tools and substitute Copilot-native equivalents where appropriate.
  4. Write or overwrite .github/agents/<name>.agent.md.
  5. Confirm: "Agent '<name>' synced to .github/agents/<name>.agent.md."

Claude-only Tool Exclusion List (sync, create, update)

When projecting the tools: array from a Claude agent definition into a Copilot agent file, the following Claude Code-specific built-in tool names MUST be removed. They do not exist in the GitHub Copilot / VS Code custom agent environment, and including them produces an invalid Copilot agent configuration.

Excluded tools — never copy to .github/agents/*.agent.md:

  • Bash
  • Glob
  • Grep
  • Read
  • TodoWrite
  • WebFetch
  • WebSearch
  • PushNotification
  • ToolSearch

Tools that are kept or translated:

  • MCP tools (any tool whose name contains mcp__ or that maps to an MCP server, e.g. mcp__docker-mcp-gateway__search, microsoft_docs_search) — keep verbatim, they are platform-portable.
  • Skill / Agent invocations — keep verbatim.
  • Translate Claude built-ins to Copilot-native equivalents where there is a 1:1 match (apply during sync only when the Copilot file currently lacks the equivalent):
    • Read → read
    • Edit / Write → edit
    • Glob / Grep → search
    • Bash → execute
    • TodoWrite → todo
    • Agent → agent

Procedure for translating tools::

  1. Read the Claude tools: array.
  2. For each tool, look it up in the Excluded list — if matched, drop it.
  3. For each remaining tool, look it up in the translation map — if matched, substitute the Copilot-native alias.
  4. Preserve all MCP and Skill/Agent entries verbatim.
  5. Emit the translated array as the Copilot file's tools: frontmatter.

Mode: deprecate-agent

Args: deprecate-agent <name> <reason>

  1. Read .claude/agents/<name>.md; fail if it does not exist.
  2. Add status: deprecated and a deprecated-reason: <reason> field to the frontmatter of .claude/agents/<name>.md.
  3. Add status: deprecated to the frontmatter of .github/agents/<name>.agent.md if it exists.
  4. Do not delete any files.
  5. Confirm: "Agent '<name>' deprecated on both platforms."

Validation Rules

  • Agent files must never be deleted — only deprecated via status: deprecated in frontmatter.
  • Claude agent files (.claude/agents/*.md) must have: name, description, tools, model.
  • Copilot agent files (.github/agents/*.agent.md) must have: name, description, tools.
  • Copilot agent tools: arrays MUST NOT contain any of the Claude-only excluded tools: Bash, Glob, Grep, Read, TodoWrite, WebFetch, WebSearch, PushNotification, ToolSearch. Reject the write if any are present.
  • Every Claude agent must have a matching Copilot agent of the same name — warn when they are out of sync.
  • Agent names must be kebab-case and match across both platform files.
  • Never read or modify .env files or sensitive configuration.

04 同仓库更多 Skills

rafsanulhasan
rafsanulhasan/HypermediaEngine

write-tests

Test implementation skill for HypermediaEngine. Takes a confirmed test plan from design-test-cases and produces compilable, runnable xUnit tests following project conventions. Validates coverage quality through mutation testing. Invoked by the sqa-engineer agent after the test plan is approved.

测试与质量
★ 0
rafsanulhasan
rafsanulhasan/HypermediaEngine

write-documentation

Structured documentation writing workflow for HypermediaEngine. Writes or updates README.md files at the repo root and in each project directory (src/, tests/, samples/, etc.) based on current implementation, ADRs, and architecture decisions.

★ 0
rafsanulhasan
rafsanulhasan/HypermediaEngine

write-adr

Writes a numbered Architecture Decision Record (ADR) for a significant architectural choice in the HypermediaEngine project. Invoked by the software-architect agent whenever a consequential design decision is made or ratified. Saves the ADR to docs/architecture/decisions/.

★ 0
rafsanulhasan
rafsanulhasan/HypermediaEngine

tunit-playwright-ui-testing

Comprehensive guidance for writing coded end-to-end UI tests in C# using TUnit.Playwright. Tests inherit from PageTest — the base class manages all browser lifecycle. Covers NuGet setup, PageTest inheritance, Page Object Model, browser configuration, parallelism limiting, and CI integration. Based on the official TUnit docs at https://tunit.dev/docs/examples/playwright.

设计师测试与质量
★ 0
rafsanulhasan
rafsanulhasan/HypermediaEngine

triage

Task classification, decomposition, dependency mapping, and routing skill. Invoked by the triage-agent to break any user request into discrete, independently-routable work items with assigned priorities and agent chains.

Skill与智能体开发
★ 0
rafsanulhasan
rafsanulhasan/HypermediaEngine

system-design

Expert system design skill. Use PROACTIVELY to design systems, review architectures, and assist software engineers in learning design principles and implementing them in the code with design patterns.

设计师视觉设计
★ 0