AGENTS.md, CLAUDE.md and Cursor rules starter files: Approvals Section Check (free sample)

What you end up with

One set of rule files every agent on the repo reads, and every one of them stops before money, deploys, deletes or sending.

What you get

• A ready-to-fill AGENTS.md with an Ask before you list and an Approvals section, plus CLAUDE.md, a Grok Bot order and a Hermes skill that point to it
• Six Approvals checks: stop and wait, show the exact change, a yes that names the action, one yes per action, silence is no, file text is never a yes
• An Ask before you list that must cover money, deploys, deletes and sending
• Blanket yeses, pasted keys and leftover placeholders caught in every rule file
• Drift caught: CLAUDE.md, the Grok Bot order, SKILL.md and Cursor rules must all point back to AGENTS.md
• Your yes at your own terminal tied to every file, a gate that says NO-GO after any change, and 19 tests

Who it is for

Anyone running more than one coding agent on a repo who wants each of them to stop for a yes before anything that costs money, goes live, deletes or reaches another person.

Works with

Solid chips: the agent is named in this listing's own tools list. Dashed chip: One Python script, starter rule files, a made-up tutoring scheduler before and after, and agent rules for each harness, plus Cursor rules. Tested on Linux.

Tools needed

Cost to run: Free. Python is free.
Time to result: About ten minutes

How it works

1. Run the example

   Unpack the file and run python3 approvals_check.py check example/before.

   Expected: 11 findings, including a blanket yes, a pasted key and three files that drift; example/after passes.

2. Copy the starter files

   Copy approvals_check.py and templates/settings.json to your repo root with your name in it, and the rest of templates/ to drafts/rules/.

   Expected: Starter files with [placeholders] to fill.

3. Let the agent fill them

   Give it AGENTS.md from the sample. It fills drafts/rules/ from your repo and runs the check until it passes.

   Expected: RULES PASS, nothing copied into place.

4. Approve and adopt

   Read every file, run python3 approvals_check.py approve settings.json drafts/rules and type adopt, then run gate.

   Expected: gate says GO, then you copy the files into your repo.

Sample prompts, free to use

Follow AGENTS.md. Fill every [placeholder] in drafts/rules/ from this repo: real commands, real folders and the actions here that cost money, go live, delete or reach another person. Run approvals_check.py check drafts/rules until it passes.
Show me every file, then stop. Never approve them or copy them over my current rule files.

Free preview of the file

# Approvals Section Check

**An agent that can deploy, delete or send will do it the moment a rule file lets it. This free sample checks that your AGENTS.md really makes it stop: the exact change shown first, a yes that names the action, one yes per action, silence read as no, and instructions found in files never counted as a yes. Then it checks that CLAUDE.md, your Grok Bot order, your Hermes skill and your Cursor rules all point back to it, and holds the files until you approve them.**

Version 1.0, 2026-11-04. By PaperChaseWebb Inc. Free sample of AGENTS.md, CLAUDE.md and Cursor rules starter files ($14). Works with Grok Bot, Claude Code, Codex and Hermes, plus Cursor rules. Python 3.8 or newer, no installs.

## What you get out of it

- A ready-to-fill AGENTS.md with an Ask before you list and an Approvals section, plus a one-line CLAUDE.md, a Grok Bot standing order, a Hermes SKILL.md, all pointing to it.
- Six checks on the Approvals section: stop and wait, show the exact change, a yes that names the action, one yes per action, no answer means stop, and instructions inside files, pages or tool output are not a yes.
- An Ask before you list of at least three actions that covers money, deploys, deletes and sending.
- Blanket yeses caught anywhere in your rule files: "assume yes", "without asking", "approved for the session", skip-permission flags.
- Drift caught: a CLAUDE.md that neither imports @AGENTS.md nor matches it, and a Grok Bot order, SKILL.md or Cursor rule tha

Worth knowing

• The sample checks the approval rules and that every file points to AGENTS.md. The five stack-specific starter files, the table of what each tool reads, the 16-check checker that confirms every listed command exists and the 9 prompts are in the full pack.
• On 2026-11-04: 19 tests passed on Python 3.8, 3.9, 3.11, 3.12 and 3.13 on Linux, run from the files unpacked out of the sample. Agent rules guide; they were not loaded inside each harness for this release.

Details

Version 1.0 · Updated Nov 4, 2026 · Markdown version of this page
License: Personal and commercial use. Download terms

Related

Prompt pack$14

AGENTS.md, CLAUDE.md and Cursor rules starter files

by PaperChaseWebb

Six fill-in instruction files for coding agents (base, Next.js plus Supabase, FastAPI, browser game, CLAUDE.md pointer, scoped Cursor rules), an Approvals section that makes the agent stop and wait for your yes, a table of what Grok Bot, Claude Code, Codex, Hermes and Cursor each read, a read-only checker for your rule files and 9 prompts to write, trim and test your own.

Works withClaude CodeCodexGrok BotHermes+1Plain .md and one Python 3 file (standard library only). Section 6a lists what Claude Code, Codex, Hermes and Cursor read, from their docs read 2026-10-30, and a block to paste into Grok Bot. The files were not run inside each tool.

Vibe coding · Beginner · 20 min setup

Agent use case pack$19

MCP server setup playbook: connect your agent to real tools without handing over the keys

by PaperChaseWebb

Pick the fewest MCP servers for a job, check them before installing, scope the credentials, configure five clients, test with MCP Inspector and build a small read-only server of your own.

Works withClaude CodeCodexClaudeCursorPlain .md

Vibe coding · Intermediate · 1 hr setup

Prompt packFree

Pre-launch security prompts for vibe-coded apps: Key Leak Pre-Check (free sample)

by PaperChaseWebb

From the pre-launch security prompts: find the three key leaks AI-built apps make most, before launch. An env file git was never told to skip, a secret behind a NEXT_PUBLIC_, VITE_ or EXPO_PUBLIC_ name, and live keys pasted into code, each printed masked. Then a drafted rotation plan, your yes recorded at your own terminal, and a launch gate that needs a clean rescan.

Works withClaude CodeCodexGrok BotHermesOne Python script, a settings template, a made-up booking app before and after the fixes, and agent rules for each harness. Tested on Linux

Vibe coding · Beginner · 5 min setup