Catch leaked secrets and risky agent actions before they run with Doorman
Pinata built Doorman, an open-source guard that runs locally next to Claude Code, Codex or Cursor. It checks what the coding agent is about to do and flags exposed API keys, suspicious instructions planted in files and risky actions before they happen.
Source
@pinatacloud on X · 2026-10-20
Written in our own words. Credit to the original author. Link checked 2026-10-21.
What they did
Pinata built Doorman, an open-source guard that runs locally next to Claude Code, Codex or Cursor. It checks what the coding agent is about to do and flags exposed API keys, suspicious instructions planted in files and risky actions before they happen.
Result (as claimed)
Released as open source; it runs on your own machine, so prompts and keys are not sent to another service.
Tools they used
Cost: Claude Code: from $17/mo; OpenAI Codex: free plan, paid from $8/mo; Cursor: free plan, paid from $8/mo.
Set it up
Pick a contained repo or task
Start on a branch or side project where mistakes are cheap.
Install and configure
Set up Claude Code, OpenAI Codex, Cursor. Read what it can run on your machine and limit permissions where possible.
Run it on a real task
Give it a well-scoped issue and let it work. Review the diff like any pull request.
Add guardrails
Keep tests and CI in the loop, and never let an agent push to main or touch production secrets unattended.
Copy the full setup prompt
Paste into your agent and it will walk you through the setup, asking before anything that sends, posts or spends.
You are helping me set up this AI use case end to end in Claude Code. Goal: Catch leaked secrets and risky agent actions before they run with Doorman What the original author did: Pinata built Doorman, an open-source guard that runs locally next to Claude Code, Codex or Cursor. It checks what the coding agent is about to do and flags exposed API keys, suspicious instructions planted in files and risky actions before they happen. Reported result (as claimed by the author): Released as open source; it runs on your own machine, so prompts and keys are not sent to another service. Source: https://x.com/pinatacloud/status/2112620850113556566 Tools they used: Claude Code, OpenAI Codex, Cursor Category: Dev & ops Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
You are helping me set up this AI use case end to end in Codex. Goal: Catch leaked secrets and risky agent actions before they run with Doorman What the original author did: Pinata built Doorman, an open-source guard that runs locally next to Claude Code, Codex or Cursor. It checks what the coding agent is about to do and flags exposed API keys, suspicious instructions planted in files and risky actions before they happen. Reported result (as claimed by the author): Released as open source; it runs on your own machine, so prompts and keys are not sent to another service. Source: https://x.com/pinatacloud/status/2112620850113556566 Tools they used: Claude Code, OpenAI Codex, Cursor Category: Dev & ops Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
You are helping me set up this AI use case end to end in Grok Bot. Goal: Catch leaked secrets and risky agent actions before they run with Doorman What the original author did: Pinata built Doorman, an open-source guard that runs locally next to Claude Code, Codex or Cursor. It checks what the coding agent is about to do and flags exposed API keys, suspicious instructions planted in files and risky actions before they happen. Reported result (as claimed by the author): Released as open source; it runs on your own machine, so prompts and keys are not sent to another service. Source: https://x.com/pinatacloud/status/2112620850113556566 Tools they used: Claude Code, OpenAI Codex, Cursor Category: Dev & ops Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
The Daily 5
Get five new use cases tomorrow morning
Free. Real sources. Unsubscribe any time.
Guides for this use case
How to run AI coding agents in a sandbox · How to vibe code an app with Claude Code · How to use Claude Code and Codex together · How to use Cursor agents and automations
Related use cases
Keep agent secrets behind an open-source credential vault
Built Agent Vault, an open-source credential proxy and vault so agents never see raw secrets.
Let your next coding agent search what the last one already tried
Switching from Claude Code to Codex or Cursor meant explaining the same work again, so a developer built Agent Recall: a local, redacted index of past agent transcripts that the next agent can search.
Point at your screen and talk to your coding agent
Point and Yap by @maurermarc lets you tap Control twice, point at something on screen and talk; Claude Code, Codex or Cursor receives it through MCP, with on-device speech-to-text. Claude co-wrote 25 of its 30 recent commits.
Audit a vibe-coded app for leaked secrets, missing RLS and dead buttons
Vibe Aegis v2.0 is an open-source safeguard and adversarial audit loop that runs in one command across Claude Code, Codex and Cursor. It targets what agents often ship: exposed NEXT_PUBLIC_ secrets, tables without row-level security, IDOR bugs, client-trusted checkout prices, double-spend race conditions and dead placeholder links.