agentleFS
Sign inStart free
tofusitotofusito/multi-agents/skills/agent-creator/SKILL.md

agent-creator

Skill0 starsChanged 6 months agoMITView on GitHub

Create or update agents for OpenCode and Claude Code with conservative permissions, cost-aware model selection, and a short questioning workflow before writing files.

  • Commits and pushes

We split this skill into the rules it gives an agent

11 rules. 0 are common, 5 are distinctive, and 6 are very specific to this project.

  • 0

    Common

    Most of the 23 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use this skill when creating or modifying an agent file (.md for OpenCode, or a custom agent for Claude Code SDK).

    line 13
  2. Distinctive:

    Before writing a new agent, collect only the information that affects safety or cost:

    line 17
  3. Distinctive:

    If the user already gave enough information, proceed without asking.

    line 27
  4. Very specific:

    Use {{MODEL_LITE}} for low-risk lookup, documentation, simple creation tasks.

    line 32
  5. Very specific:

    Use {{MODEL_STANDARD}} for normal implementation tasks that need reliability.

    line 33
  6. Very specific:

    Use {{MODEL_HEAVY}} for security, architecture, complex review, or broad code changes.

    line 34
  7. Very specific:

    Default to temperature: 0.1 for review, planning, security, and config generation.

    line 35
  8. Very specific:

    Use steps for agents that can loop or run investigations.

    line 36
  9. Distinctive:

    Do not give destructive commands automatic approval.

    line 37
  10. Distinctive:

    Keep generated agents small, explicit, and easy to audit.

    line 39
  11. Distinctive:

    Ask only the safety/cost questions that remain unanswered.

    line 89

“Common” is measured: kept by at least 6 in 10 of the 23 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

When to use me

Use this skill when creating or modifying an agent file (.md for OpenCode, or a custom agent for Claude Code SDK).

Questions first

Before writing a new agent, collect only the information that affects safety or cost:

  1. Purpose — what job should the agent do?
  2. Scope — global or this project only?
  3. Mode — primary, subagent, or all?
  4. Model tier — lite/cheap, standard, heavy, or inherit?
  5. Write access — no edits, docs/config only, or broader?
  6. Shell access — deny, ask, or allow only specific read/test commands?
  7. Network access — deny, ask, or allow specific tools?

If the user already gave enough information, proceed without asking.

Design rules

  • Split agents by risk and permission, not by personality.
  • Use {{MODEL_LITE}} for low-risk lookup, documentation, simple creation tasks.
  • Use {{MODEL_STANDARD}} for normal implementation tasks that need reliability.
  • Use {{MODEL_HEAVY}} for security, architecture, complex review, or broad code changes.
  • Default to temperature: 0.1 for review, planning, security, and config generation.
  • Use steps for agents that can loop or run investigations.
  • Do not give destructive commands automatic approval.
  • Deny or require approval for rm, sudo, git push, broad destructive patterns, and secret-reading commands.
  • Keep generated agents small, explicit, and easy to audit.

File placement

ToolGlobal pathProject path
OpenCode~/.config/opencode/agents/<name>.md.opencode/agents/<name>.md
Claude Code~/.claude/agents/<name>.md.claude/agents/<name>.md

In this repo, agents live at agents/<name>.md and are synced by distribute.sh.

Agent template

---
description: One precise sentence describing when to use this agent.
mode: subagent
hidden: true
model: {{MODEL_STANDARD}}
temperature: 0.1
steps: 10
permission:
  edit:
    "*": deny
  bash:
    "*": deny
    "git status*": allow
    "git diff*": allow
    "ls *": allow
    "rg *": allow
  webfetch: ask
  task:
    "*": deny
    "worker-lite": allow
---
You are ...

## Scope
What this agent does.

## Boundaries
What this agent must never do.

## Output
Expected output format.

Creation workflow

  1. Decide whether this is an agent, a skill, or both.
  2. Ask only the safety/cost questions that remain unanswered.
  3. Choose the cheapest model tier that can do the job reliably.
  4. Draft the frontmatter: description, mode, model, temperature, steps, permissions.
  5. Draft the prompt with concrete duties, non-goals, and expected output format.
  6. Write the file.
  7. Tell the user the invocation name and the file path.

Review checklist

Before finishing, verify:

  • Filename matches the intended agent name.
  • Description is specific enough for discovery.
  • Model choice matches risk and expected context size.
  • Edit permissions are no broader than needed.
  • Bash permissions do not auto-allow destructive, credential, or publish commands.
  • The prompt tells the agent when to ask questions and when to proceed.

Download now (ZIP)

  • Other versions25 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 25

    1. OpenHands/extensions★ 167same name
    2. 9thLevelSoftware/legion★ 76same name
    3. DGouron/review-flow★ 42same name
    4. monaccode/astromesh★ 38same name
    5. krzysztofsurdy/code-virtuoso★ 36same name
    6. different-ai/openwork-hub★ 35same name
    7. HashLips/agent-skills★ 29same name
    8. mhylle/claude-skills-collection★ 20same name
    9. WalletConnect/skills★ 14same name
    10. frankxai/Starlight-Intelligence-System★ 11same name

    15 more not listed

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • Used alongsideProjects with it also use security and security-review.

    Other files people use alongside this one.

    Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
    • 0xSeroSkill

      security

      Also in 3 projects that use this

    • SentrySkill

      security-review

      Finds exploitable application security vulnerabilities in code changes. Use for Warden security scans, appsec review, OWASP-style checks, authentication or authorization bugs, injection, XSS, SSRF, path traversal, secrets, unsafe crypto, webhook verification, open redirects, or sensitive data exposure.

      Also in 3 projects that use this

  • In this repository7 other files it gives its agents.

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

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.

Check your setup, free