Claude Prompt Library

30 Claude Code Prompts for Subagents

30 copy-paste prompts

Name the job, the tools, and the files. Claude Code writes or tightens subagents in that scope and does not invent a second team. Not a swarm and not a silent extra model.

In short: This page contains 30 copy-paste ready prompts, organized into 6 categories with a description and pro tip for each. The first 5 prompts are free instantly, no signup needed. Hand-curated and tested by the AI Academy team.

By Louis Corneloup · Founder, Techpresso

Last updated September 8, 2026·Hand-curated & tested by the AI Academy team

Add

5 prompts

One Agent

1/30

✨ What it does

Claude Code adds one subagent in [JOB] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are an agent writer working in Claude Code. <context> I need you to add one subagent I named. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [JOB] - Tools: [TOOLS] - Files: [FILES] </inputs> <task> 1. Agent in [FILES] 2. Only [TOOLS] 3. Job from [JOB] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent a second agent.

Explore Like

2/30

✨ What it does

Claude Code adds a read-only explore agent in [JOB] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are an explore writer working in Claude Code. <context> I need you to add a read-only explore agent. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [JOB] - Files: [FILES] </inputs> <task> 1. Read-only tools 2. Stay in [FILES] 3. No edit tools </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not give it Bash write.

Reviewer

3/30

✨ What it does

Claude Code adds a review subagent in [JOB] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a reviewer-agent writer working in Claude Code. <context> I need you to add a review subagent. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [JOB] - Files: [FILES] - Tools: [TOOLS] </inputs> <task> 1. Review only 2. [TOOLS] 3. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not let it merge.

Tester

4/30

✨ What it does

Claude Code adds a test-runner subagent in [TEST] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a tester-agent writer working in Claude Code. <context> I need you to add a test-runner subagent. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [TEST] - Files: [FILES] - Tools: [TOOLS] </inputs> <task> 1. Run [TEST] only 2. [TOOLS] 3. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent a runner.

Docs

5/30

✨ What it does

Claude Code adds a docs subagent in [JOB] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a docs-agent writer working in Claude Code. <context> I need you to add a docs subagent. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [JOB] - Files: [FILES] - Tools: [TOOLS] </inputs> <task> 1. Docs only 2. [TOOLS] 3. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not let it edit src unless I said.

Tighten

5 prompts

Fewer Tools

6/30

✨ What it does

Claude Code cuts agent tools in [AGENT FILE] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a tools cutter working in Claude Code. <context> I need you to cut tools on this agent. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [AGENT FILE] - Keep: [KEEP] - Files: [FILES] </inputs> <task> 1. Keep [KEEP] 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent a tool.

Narrow Job

7/30

✨ What it does

Claude Code narrows the agent job in [AGENT FILE] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a job cutter working in Claude Code. <context> I need you to narrow this agent job. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [AGENT FILE] - Job: [JOB] - Files: [FILES] </inputs> <task> 1. Job from [JOB] 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: A narrow job wanders less.

Model Honest

8/30

✨ What it does

Claude Code sets the agent model in [AGENT FILE] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a model editor working in Claude Code. <context> I need you to set the model I named. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [AGENT FILE] - Model: [MODEL OR VERIFY] - Files: [FILES] </inputs> <task> 1. Model or ask 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent opus.

Readonly

9/30

✨ What it does

Claude Code makes the agent read-only in [AGENT FILE] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a readonly editor working in Claude Code. <context> I need you to make this agent read-only. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [AGENT FILE] - Files: [FILES] </inputs> <task> 1. Strip write tools 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Read-only means read-only.

No Network

10/30

✨ What it does

Claude Code blocks agent network in [AGENT FILE] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a network cutter working in Claude Code. <context> I need you to block network on this agent. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [AGENT FILE] - Files: [FILES] </inputs> <task> 1. Strip web tools 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent a browser tool.

Use

5 prompts

When To Call

11/30

✨ What it does

Claude Code writes when to call the agent in [AGENT FILE] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a routing writer working in Claude Code. <context> I need you to write when to call this agent. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [AGENT FILE] - When: [WHEN] - Files: [FILES] </inputs> <task> 1. When from [WHEN] 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent auto-delegate.

When Not

12/30

✨ What it does

Claude Code writes when not to call in [AGENT FILE] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a stop writer working in Claude Code. <context> I need you to write when not to call it. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [AGENT FILE] - Not: [NOT] - Files: [FILES] </inputs> <task> 1. [NOT] 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: A not list saves wander.

Handoff

13/30

✨ What it does

Claude Code writes the handoff format in [AGENT FILE] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a handoff writer working in Claude Code. <context> I need you to write the handoff format. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [AGENT FILE] - Format: [FORMAT] - Files: [FILES] </inputs> <task> 1. Format from [FORMAT] 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent a JSON schema.

Parent Stays

14/30

✨ What it does

Claude Code keeps the parent in charge in [JOB] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a parent writer working in Claude Code. <context> I need you to keep the parent in charge. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [JOB] - Files: [FILES] </inputs> <task> 1. Parent plan 2. One agent call 3. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not spawn a swarm.

One At A Time

15/30

✨ What it does

Claude Code calls agents one at a time in [JOB] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a serial writer working in Claude Code. <context> I need you to call agents one at a time. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [JOB] - Files: [FILES] </inputs> <task> 1. Serial plan 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: No parallel circus unless I asked.

These prompts give you the what. Tutorials give you the why.

Learn when to use extended thinking, how to build Claude Projects, and workflows that compound. 300+ tutorials and growing.

Try AI Academy Free

Fix

5 prompts

Broken Agent

16/30

✨ What it does

Claude Code fixes the agent file in [ERROR] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are an agent fixer working in Claude Code. <context> I need you to fix this broken agent file. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [ERROR] - Files: [FILES] </inputs> <task> 1. Fix in [FILES] 2. What you ran 3. No extra agents </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent a field.

Dupes

17/30

✨ What it does

Claude Code dedupes agents in [FILES] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a dupe cutter working in Claude Code. <context> I need you to dedupe agents in scope. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [FILES] - Keep: [KEEP] </inputs> <task> 1. Keep [KEEP] 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not drop one I still need.

Rename

18/30

✨ What it does

Claude Code renames the agent in [OLD] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a rename writer working in Claude Code. <context> I need you to rename this agent as I asked. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [OLD] - New: [NEW] - Files: [FILES] </inputs> <task> 1. Rename 2. Stay in [FILES] 3. Update refs in scope </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent a better name.

Examples

19/30

✨ What it does

Claude Code adds agent examples in [AGENT FILE] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are an examples writer working in Claude Code. <context> I need you to add examples I pasted. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [AGENT FILE] - Examples: [EXAMPLES] - Files: [FILES] </inputs> <task> 1. Only [EXAMPLES] 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent a prompt.

Agent Docs

20/30

✨ What it does

Claude Code documents the agent in [AGENT FILE] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a docs writer working in Claude Code. <context> I need you to document this agent in scope. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [AGENT FILE] - Files: [FILES] </inputs> <task> 1. Docs in [FILES] 2. How to call it 3. Tools list </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent a second README.

Stop

5 prompts

Ask First

21/30

✨ What it does

Claude Code plans the agent then waits in [JOB] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a plan writer working in Claude Code. <context> I need you to plan the agent then wait. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [JOB] - Files: [FILES] </inputs> <task> 1. Plan only 2. Wait for GO </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not write until I say go.

No Swarm

22/30

✨ What it does

Claude Code adds no extra agents in [JOB] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a swarm cutter working in Claude Code. <context> I need you to do not add extra agents. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [JOB] - Files: [FILES] </inputs> <task> 1. One agent max 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: A swarm is out of scope.

Match Style

23/30

✨ What it does

Claude Code matches existing agent files in [JOB] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a style writer working in Claude Code. <context> I need you to match existing agent files. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [JOB] - Files: [FILES] </inputs> <task> 1. Match neighbors 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent YAML if we use md.

Off Limits

24/30

✨ What it does

Claude Code honors agent off limits in [JOB] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a fence writer working in Claude Code. <context> I need you to honor off limits on agents. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [JOB] - Off: [OFF] - Files: [FILES] </inputs> <task> 1. Honor [OFF] 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Off limits stays off.

No New Model

25/30

✨ What it does

Claude Code avoids a pricier model in [JOB] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a model cutter working in Claude Code. <context> I need you to do not pick a pricier model. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [JOB] - Files: [FILES] </inputs> <task> 1. Inherit unless I named a model 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent opus.

Most people use 10% of Claude. Tutorials unlock the rest.

AI Academy: 300+ hands-on tutorials on Claude, ChatGPT, Midjourney, and 50+ AI tools. New tutorials added every week.

Start Your Free Trial

Fixes

5 prompts

Undo Wander

26/30

✨ What it does

Claude Code reverts extra agent files in [DIFF] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a scope writer working in Claude Code. <context> I need you to revert extra agent files. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [DIFF] - Allowed: [FILES] </inputs> <task> 1. Revert extras 2. Keep the one job </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Neighbors stay.

Agent Plus When

27/30

✨ What it does

Claude Code adds an agent plus when-to-call in [JOB] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a combo writer working in Claude Code. <context> I need you to agent plus when-to-call. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [JOB] - Files: [FILES] </inputs> <task> 1. Agent 2. When </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: When stays short.

Agent Plus Tools

28/30

✨ What it does

Claude Code adds an agent plus tight tools in [JOB] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a combo writer working in Claude Code. <context> I need you to agent plus tight tools. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [JOB] - Tools: [TOOLS] - Files: [FILES] </inputs> <task> 1. Agent 2. Only [TOOLS] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Tools stay a list.

Cut Tools

29/30

✨ What it does

Claude Code strips dangerous agent tools in [AGENT FILE] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are a tools hunter working in Claude Code. <context> I need you to strip dangerous tools. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [AGENT FILE] - Files: [FILES] </inputs> <task> 1. Strip write/network if I said read-only 2. Stay in [FILES] </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Dangerous tools go.

Inventory

30/30

✨ What it does

Claude Code lists the repo agents in [FILES] without wandering outside the files you named. Plan if more than two files change, then tell me what you ran.

You are an inventory writer working in Claude Code. <context> I need you to list agents in this repo. Stay inside the scope I name. Do not refactor neighbors. Do not add packages I did not ask for. </context> <inputs> - Primary: [FILES] - What I want listed: [WANT] </inputs> <task> 1. List from [FILES] 2. [WANT] 3. No invented agent </task> <constraints> - Plan before you edit if more than two files change. - Do not touch files I mark OFF LIMITS. - Run the test command I gave. If I gave none, ask before inventing one. - Do not invent APIs, env vars, or tickets. </constraints> <format> A short plan, then the edits, then what you ran and what I should verify. </format>

💡

Pro tip: Do not invent a team.

Free tool

Prompt Optimizer

Turn a rough idea into a structured, professional AI prompt.

Try it free →

Frequently Asked Questions

They are written for Claude Code in the terminal: plan, edit, test, and stay inside subagents. You can paste a shorter version into claude.ai if you are not in a repo.

Each prompt sets scope and non-goals. Still use Plan mode on multi-file work, and read the diff before you accept edits.

It helps. A short CLAUDE.md with stack, test command, and files Claude must not touch keeps these prompts from inventing a second style guide.

The prompts are free to copy. Claude Code needs your Anthropic login and uses your plan's usage. Local files never leave your machine except through that session.

The hub is a 21-prompt overview. This page is only subagents: thirty copy-paste jobs with scope, verify, and non-goals already written.

Prompts are the starting line. Tutorials are the finish.

A growing library of 300+ hands-on tutorials on ChatGPT, Claude, Midjourney, and 50+ AI tools. New tutorials added every week.

Start 7-Day Free Trial Explore AI for Entrepreneurs

7-day free trial. Cancel anytime.

Browse All Claude Prompt Collections Browse All Prompt Collections

Popular prompt collections

AI Photo Editing Prompts ChatGPT Jailbreak Prompt Best ChatGPT Prompts Funny Hinge Prompts ChatGPT Caricature Prompts