一键在 Manus 中运行任何 Skill

agent-authoring

Guidance for editing Claude Code agents. Use when Claude touches agent markdown files so tool scopes, MCP wiring, memory scope, isolation, and agent hooks are configured safely.

在 Manus 中运行

概览

Guidance for editing Claude Code agents. Use when Claude touches agent markdown files so tool scopes, MCP wiring, memory scope, isolation, and agent hooks are configured safely.

安装命令

npx skills add https://github.com/mattpainter701/my-claude-setup --skill agent-authoring

复制此命令并粘贴到 Claude Code 中以安装该技能

来源

mattpainter701/my-claude-setup

星标0

分支0

更新时间2026年3月31日 19:01

SKILL.md

readonly

name	agent_authoring
description	Guidance for editing Claude Code agents. Use when Claude touches agent markdown files so tool scopes, MCP wiring, memory scope, isolation, and agent hooks are configured safely.

Agent Authoring

When editing agents:

Prefer explicit tools lists for read-only agents. Use wildcard plus disallowedTools only when you truly need the broad tool pool.
Use mcpServers to add docs/search tooling without hardcoding MCP tool names into the agent prompt.
Use initialPrompt for short steering that should apply on every spawn; keep the main prompt focused on durable review or execution behavior.
memory: project is for reusable repo knowledge. memory: local is for machine-specific workflows that should not leak across environments.
background: true is for long-running monitors or validators that do not need to block the main thread.
Agent hooks are session-scoped to the agent lifetime. Stop hooks become SubagentStop, so keep them outcome-oriented and lightweight.

同仓库更多 Skills

同仓库

hook-authoring

mattpainter701/my-claude-setup

Guidance for editing Claude Code hooks. Use when Claude touches hook scripts, hook JSON, or settings files that register hooks so matcher selection, shell choice, timeouts, and blocking behavior stay correct.

2026-03-310

memory-extraction

mattpainter701/my-claude-setup

Automatic memory extraction skill — triggered by file operations to extract durable learnings from session context. Runs in the background via the memory-extractor agent.

2026-03-310

skill-authoring

mattpainter701/my-claude-setup

Guidance for editing Claude Code skills. Use when Claude touches skill folders or SKILL.md files so slash-command naming, frontmatter, helper scripts, and conditional paths stay coherent.

2026-03-310

bom

mattpainter701/my-claude-setup

BOM (Bill of Materials) management for electronics projects — the primary orchestrator skill that coordinates DigiKey, Mouser, LCSC, element14, JLCPCB, PCBWay, and KiCad skills into a unified workflow. Create, update, and maintain BOMs with part numbers, costs, quantities stored as KiCad symbol properties. ALWAYS trigger this skill for any task involving component sourcing, pricing, ordering, distributor searches, BOM export, or fabrication preparation — even if the user names a specific distributor or fab house (e.g. "search DigiKey for...", "generate JLCPCB BOM", "order from Mouser"). This skill decides which distributor/fab skills to invoke and in what order. Also trigger on phrases like "what parts do I need", "order components", "how much will this cost", "export for JLCPCB", "find parts for this board", "cost estimate", "compare pricing", or "check stock".

2026-03-310

digikey

mattpainter701/my-claude-setup

Search DigiKey for electronic components and download datasheets — primary source for prototype orders and the preferred API-based method for fetching datasheets. Find parts by keyword or part number, check pricing/stock, download datasheets directly via API, analyze specifications. Sync and maintain a local datasheets directory for a KiCad project — extract components from schematics, download all missing datasheets, keep them up to date. Use with KiCad for BOM creation and part selection. Use this skill when the user asks about electronic components, part specifications, datasheets, footprints, pricing, stock availability, or needs to download/read a datasheet — even if they don't mention "DigiKey" by name. Also use when the user says "sync datasheets", "download datasheets for my board/project", or mentions a datasheets directory. DigiKey is the default distributor for prototyping and the preferred datasheet source because its API returns direct PDF links without web scraping. For package cross-reference t

2026-03-310

element14

mattpainter701/my-claude-setup

Search Newark, Farnell, and element14 for electronic components — find parts by MPN or distributor part number, check pricing/stock, download datasheets, analyze specifications. One unified API covers all three storefronts (Newark for US, Farnell for UK/EU, element14 for APAC). Free API key, simple query-parameter auth, no OAuth. Datasheets download directly from farnell.com CDN with no bot protection. Sync and maintain a local datasheets directory for a KiCad project. Use this skill when the user mentions Newark, Farnell, element14, needs parts from a non-US distributor, wants to compare pricing across regions, or needs datasheets from a source that doesn't require complex API auth. For package cross-reference tables and BOM workflow, see the `bom` skill.

2026-03-310

来源

mattpainter701

mattpainter701/my-claude-setup

打开 GitHub 仓库查看创作者相关仓库

安装命令

下载

在 Manus 中运行

适用职业SOC

软件开发工程师计算机与数学类职业15-1252L4

name	agent_authoring
description	Guidance for editing Claude Code agents. Use when Claude touches agent markdown files so tool scopes, MCP wiring, memory scope, isolation, and agent hooks are configured safely.

Agent Authoring

When editing agents:

Prefer explicit tools lists for read-only agents. Use wildcard plus disallowedTools only when you truly need the broad tool pool.
Use mcpServers to add docs/search tooling without hardcoding MCP tool names into the agent prompt.
Use initialPrompt for short steering that should apply on every spawn; keep the main prompt focused on durable review or execution behavior.
memory: project is for reusable repo knowledge. memory: local is for machine-specific workflows that should not leak across environments.
background: true is for long-running monitors or validators that do not need to block the main thread.
Agent hooks are session-scoped to the agent lifetime. Stop hooks become SubagentStop, so keep them outcome-oriented and lightweight.