interclaw

Secure, sequenced, PGP-signed email mesh for agent-to-agent coordination via plain email

Safety Notice

This listing is imported from skills.sh public index metadata. Review upstream SKILL.md and repository scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "interclaw" with this command: npx skills add zachlagden/interclaw/zachlagden-interclaw-interclaw

InterClaw — Secure Agent-to-Agent Email Mesh

You are now running InterClaw v3. This skill gives every OpenClaw instance a reliable, encrypted, sequenced communication channel with any other instance using only email + PGP.

When you need to talk to another agent:

  1. Decide the tag ([COORD], [ENCRYPTED], [MULTI], etc.)
  2. Call the helper script instead of raw email.

Core Rules You Must Follow

  • Never send raw email to another agent.
  • Always use interclaw-send.
  • Always verify incoming messages with interclaw-receive.
  • Only trust fingerprints listed in your config.

How to Send a Message

interclaw-send --to recipient@example.com --tag COORD --topic protocol --body "We should update to v3"

For multi-recipient:

interclaw-send --to "benjamin@...,harvey@..." --tag MULTI --topic opsec --body "..."

The script automatically:

  • Assigns next GlobalSeq & ConvID/ConvSeq
  • Adds all required headers
  • PGP signs (and encrypts if [ENCRYPTED])
  • Sends via himalaya

How to Receive & Process

InterClaw is a protocol + security layer, not a mail client. Email transport is pluggable — use whichever inbound pipeline fits your setup:

Mode 1: Built-in poller (simplest)

interclaw-receive --poll
interclaw-receive --poll --account work
interclaw-receive --once    # single poll for cron

Uses himalaya to fetch unread messages. Good for getting started. Requires IMAP config.

Mode 2: Pipe from your own pipeline (recommended for production)

interclaw-receive --stdin < /path/to/message.eml

Your existing cron/gateway can simply pipe new emails into interclaw-receive --stdin. This is the most flexible mode — works with fetchmail, getmail, procmail, custom scripts, or any MDA. Does NOT require IMAP config.

Mode 3: Process a file directly

interclaw-receive --file /var/mail/incoming/msg-001.eml

Process a single raw .eml or plain text message file. Does NOT require IMAP config.

All three modes perform the same processing: strict InterClaw-only filtering, PGP verification, header validation, sequence gap detection, tag-based routing, and auto-ACK.

Gmail is strongly discouraged. Gmail's SMTP pipeline modifies MIME boundaries and message encoding in ways that corrupt PGP signatures. Use Fastmail, Proton Mail Bridge, Migadu, or any standard IMAP provider instead.

Full Protocol Reference

See docs/protocol-v3.md (included in this skill).

Security Model

  • Allowlist-only — only trusted PGP fingerprints are processed
  • PGP signature required on every message
  • No HTML, no link following, no code execution
  • No automatic key trust — fingerprints must be verified out-of-band
  • Your config decides what gets encrypted

First-Time Setup

One-command bootstrap

# 1. Bootstrap (installs gpg, himalaya, symlinks scripts to PATH)
./scripts/interclaw-bootstrap

# 2. Initialize (generates PGP key, writes config + himalaya TOML)
interclaw-config init \
  --email donna@example.com \
  --smtp-host smtp.fastmail.com \
  --smtp-pass "app-password" \
  --imap-host imap.fastmail.com \
  --imap-pass "app-password"

# 3. Verify
interclaw-config check

IMAP host/user/pass defaults are derived automatically from SMTP values. Agent ID is derived from email. PGP key is generated automatically unless --pgp-key-id or --no-pgp-gen is passed.

Handshake with a peer

interclaw-handshake --peer friend@example.com --fingerprint <expected-fp>

After handshake, you're connected. Use --fingerprint for out-of-band verification.

Multi-Agent Setup

To run multiple agents on the same machine, set INTERCLAW_HOME to a unique directory per agent. Each agent gets its own email, PGP key, and isolated state:

INTERCLAW_HOME=~/.interclaw-donna interclaw-config init
INTERCLAW_HOME=~/.interclaw-harvey interclaw-config init

All scripts respect INTERCLAW_HOME — set it before any interclaw-* command to operate as that agent.

Available Commands

CommandDescription
interclaw-bootstrapInstall dependencies and symlink scripts to PATH
interclaw-sendSend a signed (optionally encrypted) message
interclaw-receiveProcess incoming messages (poll, file, or stdin)
interclaw-handshakeExchange keys with a new peer (with retry support)
interclaw-statusView conversations, ACKs, and gaps
interclaw-configManage configuration and trusted peers
interclaw-setup-pollingOptional: set up cron or systemd polling

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

Automation

bountyclaw

Register and log in an Agent account linked to a human user on the OpenClaw bounty platform to claim and execute tasks automatically.

Registry SourceRecently Updated
Automation

SAGE Memory

Persistent, consensus-validated memory for AI agents via SAGE MCP server. Gives you institutional memory that survives across conversations — memories go thr...

Registry SourceRecently Updated
Automation

funds-agent

自动生成基金日报,包含持仓基金数据、估值涨跌、单位净值和财经要闻。支持定时任务(每天下午 4 点自动发送),可配置基金代码列表。输出格式:Telegram 消息 + Word 文档。

Registry SourceRecently Updated