agent-builder

A guide for creating, improving, or troubleshooting Claude Code subagents, which are specialized assistants that handle particular development tasks. It covers their instructions, available tools, model choice, and delegation patterns.

In plain words
What is it for?
Use it to create an agent file, refine a subagent’s system prompt, choose its model, configure tools, or decide whether a task belongs in an agent, skill, or command.
Why use it?
It helps turn a general coding assistant into a more focused helper with clear responsibilities and controlled access. It also helps diagnose why a subagent is not being called or cannot use the tools it needs.

Skill for Claude CodeCodex

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 skills/pr-pm/prpm/agent-builder
Any agent
npx skills add pr-pm/prpm --skill agent-builder
Clone the repo
git clone --depth 1 https://github.com/pr-pm/prpm

Made for: Claude Code, Codex.

Per session 40 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 4,868 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 $0.00040 $0.04868
Opus 5 $0.00020 $0.02434
Sonnet 5 $0.00008 $0.00974
Haiku 4.5 $0.00004 $0.00487

Measured 2d ago against content hash 6327d28daa03, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agent-builder 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.

.claude/skills/agent-builder/SKILL.md · 916 lines

How it starts

The opening of the file, as written. The whole thing — 916 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Builder - Claude Code Subagent Expert

Use this skill when creating, improving, or troubleshooting Claude Code subagents. Provides expert guidance on agent design, system prompt engineering, tool configuration, and delegation patterns.

When to Use This Skill

Activate this skill when:

  • User asks to create a new subagent
  • User wants to improve an existing agent
  • User needs help with agent configuration or tool access
  • User is troubleshooting agent invocation issues
  • User wants to understand when to use agents vs skills vs commands
  • User asks about agent chaining or delegation patterns

Quick Reference

Agent File Structure

---
name: agent-name
description: When and why to use this agent
tools: Read, Write, Bash(git *)
model: sonnet
---

Your detailed system prompt defining:
- Agent role and expertise
- Problem-solving approach
- Output format expectations
- Specific constraints or requirements

File Locations

Project agents (shared with team, highest priority):

.claude/agents/my-agent.md

Personal agents (individual use, lower priority):

~/.claude/agents/my-agent.md

Plugin agents (from installed plugins):

<plugin-dir>/agents/agent-name.md

Creating Effective Subagents

Step 1: Identify the Use Case

Good candidates for subagents:

  • Complex, multi-step workflows
  • Specialized expertise (debugging, security review, data analysis)
  • Tasks requiring focused context
  • Repeatable processes with specific quality bars
  • Code review and analysis workflows

NOT good for subagents (use Skills/Commands instead):

  • Simple one-off prompts (use Slash Commands)
  • Context-aware automatic activation (use Skills)
  • Quick transformations or formatting

Step 2: Design Agent Scope

Best practices:

  • Single responsibility - Each agent does ONE thing exceptionally well
  • Clear boundaries - Define what's in/out of scope
  • Specific expertise - Don't create "general helper" agents
  • Measurable outcomes - Agent should produce concrete deliverables

Read the full file on GitHub · 916 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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 · 916 lines · 40 tokens per session scan A 6327d28daa03

Subscribe to this mod's changes

agent-builder is a skill published in the GitHub repository pr-pm/prpm (120 stars, last pushed 2mo ago), licensed MIT. It adds 40 tokens to every session and 4,868 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 2026-08-30.