Open-Sourced MCP Agent Approval Gateway
sekyr95 · reddit · 2026-07-14
The author has spun a repeatedly rewritten workflow—**having agents propose actions that require human approval before execution**—into a standalone service. ### Core Design - It functions as an **MCP server + pure REST API**. - The agent calls a tool to **propose an action**, which blocks the call. - Proposals enter an inbox where users can **approve / reject / edit** them via web, Slack, Discord, or Telegram. - Actions are only executed once a human clicks "yes." - All decisions are logged in an **audit log**. ### Target Problem It is designed for **unattended scenarios**: - Cron jobs - Multiple agents running overnight - Not agents with active sessions like Claude Code The author argues that relying on a "please confirm" prompt fails here because models will bypass it, necessitating a **true external gate**. ### Current Status - The core code is **open-source under MIT** and self-hostable - A **hosted version** is also available - It is still early stage with a small user base The author is also seeking feedback: for agents taking real-world actions, **at which layer should the human approval step reside**, and should this capability be built as an MCP rather than just REST?
More from coding & agent
- AI makes software easier to build, but it also lowers the floor on quality — paw_lean · 2026-07-21
- Fable coding run costs $6.69 for 67 lines of code in a 4-minute job — bytebot · 2026-07-21
- Cursor writes better code, but ChatGPT can still control the computer — vista8 · 2026-07-21
- Agents can remember facts, but still forget how to do the job — No_Advertising2536 · 2026-07-21
- Agent skills for project downgrade and troubleshooting tested in CLAD on LS 5.22 — stspanho · 2026-07-21
- Open-source B-roll skill turns scripts into 5-second vertical clips with Codex and Gemini — yangyi · 2026-07-21