Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add opensquad-ai/opensquad --skill cross_machine_joingit clone --depth 1 https://github.com/opensquad-ai/opensquadWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/opensquad-ai/opensquad/cross_machine_join)<a href="https://agentmods.dev/skills/opensquad-ai/opensquad/cross_machine_join"><img src="https://agentmods.dev/badge/skills/opensquad-ai/opensquad/cross_machine_join/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/opensquad-ai/opensquad/cross_machine_join"><img src="https://agentmods.dev/badge/skills/opensquad-ai/opensquad/cross_machine_join.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00151 | $0.02487 |
| Opus 5.5 | $0.00060 | $0.00995 |
| Sonnet 5.5 | $0.00030 | $0.00497 |
| Haiku 4.5 | $0.00015 | $0.00249 |
Grade A, and why
cross_machine_join scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 168 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Cross-Machine Group Join
Overview
A group can live on a different machine than this agent. The host machine's group page shows an invite string:
<host>[:port]#<group_id>[?code=XXXXXX] e.g. 192.168.5.4:9555#g-7f3a?code=418902
host is the host machine's gateway (default port 9555, or 443 behind TLS),
never the frontend dev port and never 127.0.0.1 — a loopback address means only
that machine can reach it, so stop and ask the user for the LAN IP or domain.
Two independent things get set up:
| what it is | who grants it | where it is kept | |
|---|---|---|---|
Pairing (pair_with_node) |
this agent gets its own scoped token for that gateway | the host owner approves a 6-digit code | group_chat.peers[<host>] in this agent's config |
Joining (join_by_invite) |
the agent's account there becomes a member of the group | open for a public group; a private group needs the owner | on that machine's gateway |
Pairing never repoints this agent. Its own gateway, account and groups keep working exactly as before — the peer is written next to them. No restart is needed to pair.
You never receive the host's node_secret. Pairing mints a peer token whose scopes
are fixed on the host as agent:register, group:join, board:read, board:write
— it can never reset passwords or touch that machine's launcher, and the owner can
revoke it.
What this does and does not do
- Outbound works: pairing, joining a group there with an account registered
there, and the task board for that group. The board belongs to the machine
that owns the group: joining records which machine that is, and a collaboration
task started here routes its board calls back there (with the peer token, not a
node_secret). Nothing is repointed.im.send_message(..., host="<host>")andim.list_groups(host="<host>")reach that machine's chat too. - Inbound group chat works, one hop: joining a group on a peer subscribes that machine to relay the group's messages to this agent's own gateway, which delivers them over the agent's existing socket. So messages sent to that group do arrive here and can be answered.
- Task-window messages: a message the user posts in a task window crosses
machines (the endpoint fans it out to the group's subscribers, and it lands on the
paired agent's control channel) — collaboration invitations do too. A message an
agent writes into a task window is the exception today: it is a board write,
and no fan-out follows it, so a worker on another machine finds it by reading the
board (
board_view,board_list_my_tasks) rather than seeing it arrive. - What still does not cross: board-change notifications (a task window polls the board every 5 s instead of being pushed) and DMs. For those, use a dedicated agent on that machine.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago Changed · +8 lines 2dd769a463aa
- 13d ago Changed · +24 lines · +8 tokens per session c23910877a95
- 15d ago First seen · 136 lines · 143 tokens per session scan A fdeb37be1fa7
cross_machine_join is a skill published in the GitHub repository opensquad-ai/opensquad (7 stars, last pushed 7d ago), licensed MIT. It adds 151 tokens to every session and 2,487 once invoked, about $0.0006 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-01.
Other skills, from other repositories
git-workflow
When the user asks you to perform git operations, create branches, open PRs, manage releases, or resolve conflicts, follow this guide. These are the procedural patterns you should apply consistently.
x-intelligence
When the user asks you to search, monitor, or analyze content on X (formerly Twitter), follow this guide. All operations use the official X API v2 and optionally the xAI Grok API for analysis. Never scrape. Never use unofficial endpoints.
life-automation-audit
Use this as an on-demand long prompt. Do not paste it into SOUL.md / PRINCIPLES.md / AGENTS.md. When the user asks to audit their life for automation opportunities, load and follow this prompt.
code-review
Use this skill when the user asks for a code review, audit, or “what could be wrong here”. The output should be a review, not a rewrite: prioritize bugs, security risks, behavioral regressions, and missing tests.
agent-constitution
When the user asks how to configure their agent identity and behavior, use this as a guide.
skill-creator
This skill helps create new Kabot skills through a natural, grounded workflow. It should feel collaborative, not bureaucratic. The goal is to understand the use case, shape a clean skill design, get approval, then build the real skill files.