Agent Mail

Rådgivande filreservationer och meddelanden för att koordinera flera agenter som skriver till ett repo

Fungerar med konfiguration ★ 7.6/10

Vad den gör

Agent Mail är en Claude/Codex-färdighet som fungerar som en adapter över en extern "Agent Mail" backend (MCP tools eller en `am` CLI) för att koordinera flera agenter som skriver till samma repository. Den registrerar agent identities, sends threaded messages/acknowledgements, och places advisory, TTL-bounded file reservations so cooperating writers avoid clobbering each other's paths. Den utlöses av multi-agent coordination requests like "coordinate writers" or "reserve files".

Testrapport

SKILL.md hittades på skills/agent-mail/SKILL.md (mirrored in skills-codex/ and images/gemini/skills/). Frontmatter parsas med name+description; alla tre refererade filer (references/TOOLS.md, WORKFLOWS.md, RECOVERY.md) returnerade HTTP 200; sökvägar är relative with no hardcoded machine paths. Inga security smells (no curl|sh, base64, or exfiltration); den beskriver en destructive `clear-and-reset-everything` och ett git pre-commit guard men gates both behind explicit per-operation caller authorization. TRIGGER: Jag bedömde 5 phrasings — SKULLE: "coordinate two agents editing the same repo so they don't clobber files", "reserve src/auth/** while I refactor so the other writer stays off it", "set up file reservations for our multi-agent session"; SKULLE INTE: "email my client about the invoice", "reserve a conference room for the team meeting" — alla 5 decided correctly (the description clearly scopes it to agent-to-agent coordination, not real mail). OUTPUT unmeasurable: the skill is a thin adapter whose every action requires the Agent Mail MCP tools or an `am` CLI, and `which am` returned not found with no MCP present; jag kunde only fabricate ids/reservations, which would not honestly measure the skill. DOCS solid and refreshingly non-overselling (explicit "does not do" boundary, named failure modes), docked 1 point because nothing in the skill or its references tells you how to obtain/install the backend the whole skill depends on. Note: frontmatter sets user-invocable:false (meant to be consumed by the agent-native skill), so a human won't invoke it directly.

Testad: 2026-07-21 · Claude Code 2.x (agent harness)

Installation

git clone --depth 1 https://github.com/boshu2/agentops.git /tmp/agent-mail-src
mkdir -p ~/.claude/skills
cp -R /tmp/agent-mail-src/skills/agent-mail ~/.claude/skills/agent-mail
# REQUIRES an external "Agent Mail" backend that is NOT installed by the copy above:
# either the Agent Mail MCP tools (ensure_project, register_agent, send_message,
# file_reservation_paths, ...) registered in your client, OR a self-describing `am` CLI
# on PATH (discover syntax via `am mail --help`, `am file_reservations --help`).
# Neither ships in this skill folder; the skill only orchestrates them. Without one of
# them present the skill's own "Adapter unavailable" branch says to stop.
# `am`/agentops CLI appears to live in the repo's cli/ (Go); building/running it plus its
# datastore is a separate setup step not covered by the skill files.

Kommandon och exempelprompter

/agent-mailRådgivande filreservationer och meddelanden för att koordinera flera agenter som skriver till ett repo

Skills triggas av vanliga förfrågningar — inga kommandon att memorera. Efter installationen aktiverar prompter som dessa skillen (på engelska):

Coordinate these two agents using Agent Mail
Set up a lock so agents don't collide on this file
Hand off this task to another agent via inbox

Läs hela rapporten →