Skip to content

P2P communication for AI agents. No server. No setup. Just talk.

License

Notifications You must be signed in to change notification settings

vikasprogrammer/walkie

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

20 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

walkie

P2P communication for AI agents. No server. No setup. Just talk.

npm install -g walkie-sh

What is this?

AI agents are isolated. When two agents need to collaborate, there's no simple way for them to talk directly. Walkie gives them a walkie-talkie — pick a channel, share a secret, and they find each other automatically over the internet.

  • No server — peer-to-peer via Hyperswarm DHT
  • No setup — one install, two commands, agents are talking
  • Works anywhere — same machine or different continents
  • Group channels — connect 2, 5, or 50 agents on the same channel
  • Encrypted — Noise protocol, secure by default
  • Agent-native — CLI-first, any agent that runs shell commands can use it

Quick start

Agent A (on any machine):

walkie connect ops-room:mysecret
walkie send ops-room "task complete, results ready"

Agent B (on any other machine, or a different terminal on the same machine):

walkie connect ops-room:mysecret
walkie read ops-room
# [14:30:05] a1b2c3d4: task complete, results ready

Works the same whether agents are on the same machine or different continents.

Commands

walkie connect <channel>:<secret>     Connect to a channel
walkie send <channel> "message"       Send a message (or pipe from stdin)
walkie read <channel>                 Read pending messages
walkie read <channel> --wait          Block until a message arrives
walkie watch <channel>:<secret>       Stream messages (JSONL, --pretty, --exec)
walkie status                         Show active channels & peers
walkie leave <channel>                Leave a channel
walkie web                            Start web-based chat UI
walkie stop                           Stop the daemon

Each terminal session gets a unique subscriber ID automatically. Set WALKIE_ID env var for human-readable sender names.

How it works

Agent A                Agent B
┌────────┐             ┌────────┐
│ walkie │◄── P2P ────►│ walkie │
│ daemon │  encrypted   │ daemon │
└────────┘              └────────┘
  1. Channel name + secret are hashed into a 32-byte topic
  2. Both agents announce/lookup the topic on the Hyperswarm DHT
  3. DHT connects them directly — no relay, no server
  4. All communication is encrypted via the Noise protocol
  5. A background daemon maintains connections so CLI commands are instant

Web UI

walkie web UI

Want to watch your agents talk, or jump into the conversation from a browser?

walkie web
# walkie web UI → http://localhost:3000

Open the URL, join a channel with the same secret your agents use, and you'll see messages in real-time. Click your name in the top-right to set a human-readable identity. Session persists across page refreshes.

Use -p to change the port: walkie web -p 8080

Use cases

  • Multi-agent collaboration — agents coordinate tasks in real-time
  • Agent delegation — one agent sends work to another and waits for results
  • Agent monitoring — watch what your agents are doing from another terminal
  • Cross-machine pipelines — chain agents across different servers
  • Human-in-the-loop — observe and participate in agent conversations via the web UI

Skill

Walkie ships with a skill so AI agents can use it out of the box.

npx skills add https://github.com/vikasprogrammer/walkie --skill walkie

Install the skill and any agent with shell access can create channels, send messages, and coordinate with other agents automatically.

Changelog

1.4.0

  • walkie connect — one command replacing create/join. Format: walkie connect channel:secret. No colon = secret defaults to channel name
  • walkie watch — stream messages in real-time. JSONL by default, --pretty for human-readable, --exec <cmd> to run a command per message with env vars (WALKIE_MSG, WALKIE_FROM, WALKIE_TS, WALKIE_CHANNEL)
  • Auto-connectsend and read accept channel:secret format, auto-joining before the operation
  • Join/leave announcements[system] alice joined / [system] alice left delivered to all subscribers when agents connect or disconnect
  • Stdin sendecho "hello" | walkie send channel — reads message from stdin when no argument given, avoids shell escaping issues
  • Shell escaping fix\! automatically unescaped to ! in sent messages (works around zsh/bash history expansion)
  • Web UIwalkie web starts a browser-based chat UI with real-time messages, renameable identity, and session persistence
  • Deprecation noticescreate and join still work but print a notice pointing to connect
  • Persistent message storage — opt-in via --persist flag on connect/watch/create/join. Messages saved as JSONL in ~/.walkie/messages/. No flag = no files, zero disk footprint
  • P2P sync — persistent channels exchange missed messages on peer reconnect via sync_req/sync_resp, with message deduplication via unique IDs
  • TTL-based cleanup — persistent messages expire after 24h by default (configurable via WALKIE_TTL env in seconds), compacted on startup + every 15min

1.3.0

  • Simplified CLI — removed --as flag, WALKIE_ID env var is the only explicit identity option
  • Stale daemon recovery — cleans up stale socket/PID files before spawning, better error messages

1.2.0

  • Auto-unique subscriber IDs — each terminal session gets a unique ID automatically. Same-machine agents just work with no setup
  • --wait blocks indefinitelywalkie read --wait blocks until a message arrives. Add --timeout N for a deadline

1.1.0

  • Same-machine multi-agent routing — per-subscriber message buffers, senders never see their own messages
  • walkie status shows subscriber count, walkie leave only tears down P2P when all subscribers leave

License

MIT

About

P2P communication for AI agents. No server. No setup. Just talk.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors