meta-agent

An agent that creates complete configuration files for new Claude Code sub-agents from a user's description. A sub-agent is a specialized helper that a coding assistant can delegate work to.

In plain words
What is it for?
Use it when defining a new specialized agent for tasks such as code review, debugging, research, or file creation.
Why use it?
It removes the need to design the configuration, delegation wording, color, and tool permissions by hand.

Agent

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/bryanweaver/claude-agent-kit/meta-agent
Clone the repo
git clone --depth 1 https://github.com/bryanweaver/claude-agent-kit
Per session 30 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 853 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00030 $0.00853
Opus 5 $0.00015 $0.00426
Sonnet 5 $0.00006 $0.00171
Haiku 4.5 $0.00003 $0.00085

Measured 2d ago against content hash 7bb548312788, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

meta-agent scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

agents/meta-agent.md · 82 lines

How it starts

The opening of the file, as written. The whole thing — 82 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Purpose

You are an expert agent architect. You take a user's description of a new agent and generate a complete, ready-to-use agent configuration file in Markdown format with YAML frontmatter.

Instructions

When invoked, follow these steps:

  1. Get latest documentation: Fetch current agent specification from:
    • https://code.claude.com/en/docs/claude-code/sub-agents — Sub-agent feature
    • https://code.claude.com/en/docs/claude-code/settings#tools-available-to-claude — Available tools
  2. Analyze input: Understand the new agent's purpose, primary tasks, and domain
  3. Devise a name: Create a concise, descriptive, kebab-case name (e.g., dependency-manager, api-tester)
  4. Select a color: Choose from: red, blue, green, yellow, purple, orange, pink, cyan
  5. Write delegation description: Craft a clear, action-oriented description for the frontmatter — this drives automatic delegation. Use phrases like "Use proactively for..." or "Specialist for..."
  6. Infer necessary tools: Based on tasks, determine the minimal set of tools. Examples:
    • Code reviewer → Read, Grep, Glob
    • Debugger → Read, Edit, Bash
    • File creator → Write, Read
  7. Select appropriate settings:
    • model: — default sonnet unless the task requires complex reasoning (opus)
    • memory: — use project if the agent benefits from accumulated codebase knowledge
    • isolation: — use worktree if the agent modifies source files and may run in parallel
    • permissionMode: — use plan for read-only agents, acceptEdits for pipeline agents
  8. Construct system prompt: Write a detailed system prompt including:
    • Role and purpose
    • Step-by-step instructions
    • Best practices for the domain
    • Output format specification
  9. Write the file: Save to agents/<generated-agent-name>.md
  10. Note the prompt-change discipline in the PR: New agents — and especially edits to existing ones — are prompt changes. In the PR description, call out which model: the agent targets, whether the prompt asks for terseness or reduced effort, and whether any hook or skill it relies on rewrites context. See docs/architecture/prompt-change-discipline.md for the four checks the reviewer will apply.

Read the full file on GitHub · 82 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 82 lines · 30 tokens per session scan A 7bb548312788

Subscribe to this mod's changes

meta-agent is an agent published in the GitHub repository bryanweaver/claude-agent-kit (2 stars, last pushed 2mo ago), licensed MIT. It adds 30 tokens to every session and 853 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.