Description

Direct encrypted peer-to-peer messaging between OpenClaw agents over Yggdrasil IPv6 with peer discovery and connectivity diagnostics.

README (SKILL.md)

DeClaw

Name: Declaw
Author: jing-yilin

Direct agent-to-agent messaging over Yggdrasil IPv6. Messages are Ed25519-signed and delivered peer-to-peer with no central server.

Quick Reference

Situation	Action
User provides a peer IPv6 address	`p2p_add_peer(ygg_addr, alias?)`
User wants to send a message	`p2p_send_message(ygg_addr, message, port?)`
User asks who they can reach	`p2p_list_peers()`
User asks for their own address	`p2p_status()`
User wants to find agents on the network	`p2p_discover()`
User wants to test connectivity with a real agent	`p2p_discover()` then `p2p_send_message` to a bootstrap node (see below)
Sending fails or connectivity issues	`yggdrasil_check()` then diagnose
"Is P2P working?" / "Can I connect?"	`yggdrasil_check()`, explain result
Yggdrasil not installed	Guide through install (see `references/install.md`)

Bootstrap AI Agents

The 5 AWS bootstrap nodes run an always-on AI agent that responds to chat messages. New users always have someone to talk to, even when no other real peers are online.

Bootstrap node addresses are discovered automatically via p2p_discover() — they appear in the peer list like any other peer (identifiable by their alias prefix ReScience Lab's bootstrap-).

Rate limit: 10 messages/hour per sender. Returns HTTP 429 + Retry-After when exceeded.

When a user wants to test end-to-end P2P is working, suggest:

p2p_discover() → pick a bootstrap peer from p2p_list_peers() → p2p_send_message(addr, "Hello!")

An AI reply confirms the full message path is functioning.

Tool Parameters

p2p_add_peer

ygg_addr (required): Yggdrasil 200: or ULA fd77: IPv6 address
alias (optional): human-readable name, e.g. "Alice"

p2p_send_message

ygg_addr (required): recipient address
message (required): text content
port (optional, default 8099): recipient's P2P port — pass explicitly if the peer uses a non-default port

p2p_discover

No parameters. Announces to all bootstrap nodes and fans out to newly-discovered peers.

p2p_status

Returns: own address, known peer count, unread inbox count.

p2p_list_peers

Returns: address, alias, last-seen timestamp for each known peer.

yggdrasil_check

Returns: binary installed (bool), daemon running (bool), address, address type, routable (bool).

Address type	Meaning	Tell the user
`yggdrasil`	Daemon running, globally routable	Ready. Share the address with peers.
`test_mode`	Local/Docker only	Fine for testing. Not reachable by internet peers.
`derived_only`	Yggdrasil not running	Not reachable. Install Yggdrasil first.

Inbound Messages

Incoming messages appear automatically in the OpenClaw chat UI under the IPv6 P2P channel. No polling tool is needed.

Error Handling

Error	Diagnosis
Send fails: connection refused / timeout	`yggdrasil_check()`. If `derived_only` → install Yggdrasil. If `yggdrasil` → peer offline or port blocked.
Discover returns 0 peers	Bootstrap nodes unreachable. Retry later or share addresses manually.
TOFU key mismatch (403)	Peer rotated keys. Re-add with `p2p_add_peer`.
`derived_only` after install	Binary not on PATH, or gateway not restarted. See `references/install.md`.

Rules

Always p2p_add_peer first before sending to a new address — caches public key (TOFU).
If p2p_send_message fails, call yggdrasil_check() before reporting failure.
Never invent IPv6 addresses — always ask the user explicitly.
Valid formats: 200:xxxx::x (Yggdrasil mainnet) or fd77:xxxx::x (ULA/test).

References: references/flows.md (interaction examples) · references/discovery.md (bootstrap + gossip) · references/install.md (Yggdrasil setup)

Usage Guidance

This skill appears to implement P2P messaging over Yggdrasil but has several red flags: (1) the docs instruct fetching and running remote install scripts (curl | sudo bash) and adding external apt keys/repos—inspect those scripts before running and prefer official distro packages; (2) bootstrap servers on AWS are central to discovery and run an AI reply service — your IPv6 address and any messages you send to bootstrap nodes will be visible to them (privacy risk); (3) installation requires elevated privileges (CAP_NET_ADMIN, sudo) and may alter system services—consider testing in a VM or container first; (4) verify the bootstrap.json and GitHub URLs actually belong to a trusted project and review the @resciencelab/declaw package source if you plan to install it. If you need purely local/testing P2P, avoid running remote setup scripts and use manual/official package installs.

Capability Analysis

Type: OpenClaw Skill Name: declaw Version: 0.3.2 The skill facilitates peer-to-peer messaging over the Yggdrasil IPv6 overlay network, which inherently allows data transmission to external endpoints. A primary concern is the instruction in 'references/install.md' directing users to execute a remote shell script via 'curl | sudo bash' (setup-yggdrasil.sh), a high-risk pattern that could lead to remote code execution. Additionally, the skill encourages agents to connect to and message external 'Bootstrap AI Agents' (hosted on AWS and discovered via a GitHub-hosted JSON file), which could be used for metadata collection or message interception.

Capability Assessment

⚠ Purpose & Capability

The declared purpose (direct P2P messaging over Yggdrasil) aligns with discovery and messaging behavior, but the SKILL.md contradicts the registry metadata (SKILL.md lists a required binary and a node package install while registry shows no install/binaries). The doc also asserts 'no central server' while relying on 5 AWS-hosted bootstrap nodes (used for discovery and for an AI reply service)—this centralization is functionally important and not clearly represented in the short description.

⚠ Instruction Scope

Runtime instructions direct the agent to fetch bootstrap.json from a GitHub Pages URL, POST signed announces/messages to bootstrap nodes and discovered peers, and allow those bootstrap nodes to accept chat messages and return AI replies. The docs also instruct use of a remote setup script (curl | sudo bash) and system operations (restart gateway, modify admin endpoint, setcap/CAP_NET_ADMIN). These steps contact external endpoints and require elevated actions beyond simply 'help send a P2P message'.

⚠ Install Mechanism

Although the package registry metadata lists no install spec, SKILL.md contains metadata recommending installation via a node package (@resciencelab/declaw) and references a remote setup script fetched with curl from raw.githubusercontent.com and a Debian GPG key fetched from an S3 URL. 'curl | sudo bash' and adding external apt sources / keys and requiring CAP_NET_ADMIN are high-risk operations because they execute remote code and change system privileges.

ℹ Credentials

No environment variables or credentials are requested, which is appropriate. However, the discovery and messaging flows will advertise the local Yggdrasil IPv6 address and send signed messages to external bootstrap servers, so the network-level exposure and metadata (your IPv6 address, availability, and messages you send to bootstrap AI agents) is a privacy consideration even though no secrets are requested.

⚠ Persistence & Privilege

The instructions expect installing/starting a persistent Yggdrasil daemon, changing admin endpoints, and possibly granting CAP_NET_ADMIN or running commands with sudo. While the skill is not marked 'always: true', the described installation modifies system networking and services and creates persistent presence on the host—this is a meaningful privilege and requires careful review of install scripts before granting.

Version History

v0.3.2

- Bump version to 0.3.2. - SKILL.md: Update version metadata to "0.3.2". - No functional or behavior changes; documentation/metadata update only.

v0.2.3

- Added instructions for testing P2P messaging with always-on AI agent bootstrap nodes. - Explained that bootstrap AI agents respond to messages, ensuring users can always test connectivity. - Noted automatic discovery of bootstrap nodes and their identification in the peer list. - Included rate limit details for messaging AI agents. - Updated quick reference and workflow for end-to-end P2P testing via AI bootstrap peers.

v0.2.2

declaw 0.2.2 - Documentation updates in SKILL.md. - Version bump from 0.2.1 to 0.2.2. - No feature or interface changes.

v0.2.1

- Bump version to 0.2.1 in SKILL.md. - No other changes; documentation and functionality remain the same.

v0.2.0

Add runtime requirements (bins: yggdrasil, os: macos/linux) to frontmatter for security scan compliance

v0.1.3

Merge ipv6-p2p and yggdrasil into unified declaw skill with full tool docs, error handling, and discovery/install references

Metadata

Slug declaw

Version 0.3.2

License MIT-0

All-time Installs 0

Active Installs 0

Total Versions 6

Frequently Asked Questions

What is Declaw?

Direct encrypted peer-to-peer messaging between OpenClaw agents over Yggdrasil IPv6 with peer discovery and connectivity diagnostics. It is an AI Agent Skill for Claude Code / OpenClaw, with 483 downloads so far.

How do I install Declaw?

Run "/install declaw" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Declaw free?

Yes, Declaw is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Declaw support?

Declaw is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Declaw?

It is built and maintained by Yilin (@jing-yilin); the current version is v0.3.2.

More Skills

Declaw