Claimidx
MCP server and CLI: replay-gated index of software failures other agents already solved (ask before retrying,...
CCCXGBCUPi
SkillMaxx / Development / agent-postbox
agent-postbox
MCP server and CLI for verifiable agent handoffs over GitHub Issues and PRs: read-back on every write, receipts, no self-review, signed evidence log. Community request via Bluesky (@aethelgardshuman), Nov 2 2026.
- Stars
- ★ 0
- Last updated
- Hands-on tests
- Tested ✓ in 5 of 5
Install
Tested ✓✓ Verified from docsOfficial docs ↗
claude mcp add agent-postbox -- npx -y github:Aethelgard-Systems/agent-postbox serveTested ✓ on 2026-11-14: installed and loaded in a throwaway sandbox with no credentials.
Log excerpt
Added stdio MCP server agentpostbox with command: npx -y github:Aethelgard-Systems/agent-postbox serve to local config File modified: /home/sbx/.claude.json [project: /home/sbx] $ verify: agentpostbox: npx -y github:Aethelgard-Systems/agent-postbox serve - ✔ Connected
8 runs, latest 2026-11-14
Tested ✓✓ Verified from docsOfficial docs ↗
codex mcp add agent-postbox -- npx -y github:Aethelgard-Systems/agent-postbox serveTested ✓ on 2026-11-14: installed and loaded in a throwaway sandbox with no credentials.
Tested from a local mirror of the official openai/plugins marketplace, because the CLI reserves the official name for signed-in users.
Log excerpt
Added global MCP server 'agentpostbox'. $ verify: Name Command Args Env Cwd Status Auth agentpostbox npx -y github:Aethelgard-Systems/agent-postbox serve - - enabled Unsupported PROBE OK 9 tools (protocol ok)
8 runs, latest 2026-11-14
Tested ✓
grok mcp add agent-postbox -- npx -y github:Aethelgard-Systems/agent-postbox serveTested ✓ on 2026-11-14: installed and loaded in a throwaway sandbox with no credentials.
Log excerpt
Added stdio MCP server 'agentpostbox' with command: npx -y github:Aethelgard-Systems/agent-postbox serve to user config
File modified: ~/.grok/config.toml
$ verify:
agentpostbox: npx -y github:Aethelgard-Systems/agent-postbox serve
agentpostbox (stdio: npx -y github:Aethelgard-Systems/agent-postbox serve)
✓ handshake OK (protocol 2025-06-18)
✓ 9 tools discovered8 runs, latest 2026-11-14
Tested ✓✓ Verified from docsOfficial docs ↗
~/.cursor/mcp.json: {"mcpServers":{"agent-postbox":{"command":"npx","args":["-y","github:Aethelgard-Systems/agent-postbox","serve"]}}}Tested ✓ on 2026-11-14: installed and loaded in a throwaway sandbox with no credentials.
Log excerpt
✓ Enabled and approved MCP server: agentpostbox $ verify: agentpostbox: ready Tools for agentpostbox (9):
8 runs, latest 2026-11-14
Tested ✓
pi mcp add agent-postbox -- npx -y github:Aethelgard-Systems/agent-postbox serveTested ✓ on 2026-11-14: installed and loaded in a throwaway sandbox with no credentials.
Pi has no plugin system. This checks that Pi installs the repo as a package and finds its skills. Hooks don't load in Pi. MCP servers are tested separately with Pi's built-in MCP.
Log excerpt
Added global MCP server "agentpostbox" in /home/sbx/.pi/agent/mcp.json. Check it with: pi mcp list $ verify: agentpostbox: connected, 9 tools (codemode, global) npx -y github:Aethelgard-Systems/agent-postbox serve tools: inbox, read, comment, label, propose, review, handoff_create, handoff_read, handoff_ack
8 runs, latest 2026-11-14
README
View on GitHub ↗<p align="center"><img src="https://raw.githubusercontent.com/Aethelgard-Systems/agent-postbox/main/docs/assets/agent-postbox-banner.png" alt="agent-postbox – AI agent handoffs with receipts: Agent A hands work through the postbox to Agent B, every write is read back, self-review is blocked, a human approves the merge." width="100%"></p>
agent-postbox
Verifiable handoffs for MCP agents. agent-postbox makes agent handoffs over GitHub verifiable – with read-back and evidence instead of bare status messages.
AI coworkers collaborate through GitHub Issues and pull requests – with bounded writes, receipts and four-eyes review.
You already have a durable, auditable, human-readable coordination system: your GitHub repository. agent-postbox turns it into a mailbox for AI agents, so they can hand work to each other, report results and review each other's pull requests, while every write stays small, verified and attributable, and every decision stays with a human.
It is a small Model Context Protocol server (stdio or Streamable HTTP) plus a CLI. Zero runtime dependencies.
human ──opens issue──▶ #12 [to:builder]
builder ── inbox ─────▶ claims (claimed:builder), works, opens PR #13 [to:reviewer]
reviewer ─ review ────▶ REQUEST_CHANGES on #13 @ head commit (cannot review its own PR)
builder ── comment ───▶ evidence: tests, output, links
human ───────────────▶ merges. Agents cannot merge, close, or set authority labels.
A shared inbox delivers a message. agent-postbox proves the handoff happened – read-back of every write, an acknowledgement bound to the exact text the agent read, no self-review, and a signed evidence log you can verify.
Release: v0.2.1 · For: developers running several coding agents (Claude Code, Codex, Cursor, …) on one GitHub repo · Zero runtime dependencies.
Try it in 30 seconds
No GitHub account, token or network needed – the demo runs two agents against an in-memory GitHub with synthetic data:
git clone https://github.com/Aethelgard-Systems/agent-postbox.git
cd agent-postbox
npm run demo
Windows PowerShell blocks npm with “running scripts is disabled on this system”? Use npm.cmd run demo or node examples/offline-demo/demo.mjs instead – same demo, no policy change needed.
Preview from the source repo. Images are removed.
Similar alternatives
Get the weekly install scorecard
What broke, what's flaky, what's safe to install across Claude Code, Codex, Grok Build, Cursor and Pi. Free, every Tuesday.
Loading the signup form… or subscribe on Beehiiv
Form not loading? Subscribe on Beehiiv →