ai-extender / references
AGENTS.md0 starsChanged 39 days agoMITView on GitHub
Location: agents/<name>.md. In plugin.json, agents takes .md file paths only (not directories) and replaces the default scan. Plugin agents ignore permissionMode, mcpServers, hooks, initialPrompt. Need those → tell the user to copy the agent to .claude/agents/, or put the behavior in a skill/hook instead. The subagent sees only this plus environment basics (not Claude Code's prompt). Write: role (one line), responsibilities, process, output format, stop condition. Isolated/noisy context, hard tool restriction, parallel work, or a cheaper model for a narrow…
In this repository9 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .claude/CLAUDE.mdCLAUDE.md
- ai-extenderSkill
- claude-developerSkill
- claude-maintainerSkill
- claude-packagerSkill
- claude-plannerSkill
- claude-reviewerSkill
- prompt-builderSkill
The file
Files
Agents (subagents)
Location: agents/<name>.md. In plugin.json, agents takes .md file paths only (not directories) and replaces the default scan.
Frontmatter (camelCase; unknown keys ignored silently)
| Field | Notes |
name (req) | unique, role slug; no : and no leading - or the file is skipped |
description (req) | when Claude should delegate; add 1 or 2 concrete trigger examples |
tools / disallowedTools | least privilege; omit = inherits everything |
model | sonnet, opus, haiku, fable, full id, or inherit |
maxTurns, effort, color, background, isolation: worktree, memory, skills (preload), omitClaudeMd | optional |
color | red, blue, green, yellow, purple, orange, pink, cyan |
Plugin agents ignore permissionMode, mcpServers, hooks, initialPrompt. Need those → tell the user to copy the agent to .claude/agents/, or put the behavior in a skill/hook instead.
Body = the whole system prompt
The subagent sees only this plus environment basics (not Claude Code's prompt). Write: role (one line), responsibilities, process, output format, stop condition.
When to add one
Isolated/noisy context, hard tool restriction, parallel work, or a cheaper model for a narrow job. Otherwise a skill is cheaper. One responsibility per agent; end with a scratchpad (ai-extender/references/scratchpad.md).
Validate a folder: claude plugin validate <dir>.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.