Agent skill

Group

by xhluca in xhluca/agent-talk

Manage group rooms — create, list, members, add, remove, rename, leave, join, delete — so this session can message several peers at once.

MITAuto-check passed

Install Group

skills CLI
$ npx skills add xhluca/agent-talk --skill group -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install xhluca/agent-talk group --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/xhluca/agent-talk.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/group .claude/skills/group && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
group
GitHub stars
190
Token cost
~2k tokens
SKILL.md length
939 words
Files
1
Skills in repo
14
Repo updated
First seen
Licence
MIT

At a glance

Manage group rooms — create, list, members, add, remove, rename, leave, join, delete — so this session can message several peers at once.

  • Then send to it with the send skill (send --group NAME)
  • SKILL.md covers The room's name is just your…, Membership is cooperative — no…, Create a room and List / inspect, plus 5 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Group is an agent skill from xhluca/agent-talk. Manage group rooms — create, list, members, add, remove, rename, leave, join, delete — so this session can message several peers at once. A group is a local roster of saved peers; sending to it delivers a private copy to each member. Use to set up or adjust a room, then send to it with the send skill (send --group NAME). <user is this session's user directory (absolute path; from init).

Its SKILL.md is about 2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

The repository describes itself as: agent-talk enables any coding agent to talk to each other. The licence is MIT.

When your agent uses it

  • Then send to it with the send skill (send --group NAME)

Example prompts

  • “/group”

What it can do on your machine

Read from SKILL.md and the folder at commit 8a7fc4f. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Group loads about 2k tokens when it runs. Until then it costs about 100 tokens; SKILL.md has 939 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~100
When it runs · the whole SKILL.md, loaded when a task matches
~2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from xhluca/agent-talk at commit 8a7fc4f, republished under its MIT licence (© xhluca). 939 words, ~1,955 tokens.

Download SKILL.mdSave it as .claude/skills/group/SKILL.md (or your agent's skills folder).
name
group
description
Manage group rooms — create, list, members, add, remove, rename, leave, join, delete — so this session can message several peers at once. A group is a local roster of saved peers; sending to it delivers a private copy to each member. Use to set up or adjust a room, then send to it with the send skill (`send --group NAME`). `<user>` is this session's user directory (absolute path; from init).

group — manage a group room

A group is a local roster of peers with a friendly name. Sending to it delivers a separate end-to-end-encrypted copy to each member, so a "group" is just a convenient way to reach several saved peers at once — there is no shared server-side room, and the relay never learns who is in it.

retalk group create NAME --members <peer>,<peer> --dir "<user>/identity"   # make a room
retalk group list --dir "<user>/identity"                                  # rooms you're in
retalk group members NAME --dir "<user>/identity"                          # who's in one
retalk group add NAME <peer>,<peer> --dir "<user>/identity"                # add people
retalk group remove NAME <peer> --dir "<user>/identity"                    # take someone out
retalk group rename NAME NEW-NAME --dir "<user>/identity"                   # relabel it (local only)
retalk group leave NAME --dir "<user>/identity"                            # leave the room
retalk group join NAME --dir "<user>/identity"                             # undo a leave
retalk group delete NAME --dir "<user>/identity"                           # forget it locally

Members are your saved peers (by local name) or raw 32-hex fingerprints — the same values you'd pass to send. A member you haven't saved yet: add them first (or pass their fingerprint directly). Target the identity inline with --dir "<user>/identity"; group commands are local bookkeeping and never need the passphrase or the relay, except leave, which messages the other members (see below).

The room's name is just your label

Every room has a stable 32-hex group id that never changes, plus a name that is your local label for it — like a contact name. Renaming changes only what you call it; the other members keep their own names for the same room, and the id stays put. Two rooms can't share a name on your side (retalk refuses a duplicate), and a name that looks like a fingerprint is rejected — pick a human name.

You don't have to create a room to be in one. When a peer sends to a group that includes you, the room appears on its own the first time their message arrives (with the name they gave it), and you can reply to everyone from there.

Membership is cooperative — no owner, no admin

There's no group owner and no server-side membership list. Each member keeps their own roster, and the most recent sender's roster wins: when you add or remove someone, that change reaches the others the next time you send to the group. So "remove carol" means carol stops getting your copies right away, and stops getting everyone else's once each of them has sent once after your change. Tell the user in those terms — "carol's out on my side; she'll drop off for the others as the room keeps talking" — not as an instant server-side kick.

Create a room

retalk group create team --members bob,carol --dir "<user>/identity"
# stdout: {"group_id","name","members"}   (members = fingerprints, you included on send)

Needs at least one member. The room counts you plus the members; a relay may cap the size, and an over-cap create is refused with the limit — trim the roster and retry. After creating, tell the user the room is ready and that they can message it with the send skill (send --group team "..."); don't dump the JSON.

List / inspect

retalk group list --dir "<user>/identity"          # NAME  <id>  N member(s), one per line
retalk group list --json --dir "<user>/identity"   # {"group_id","name","members"} per line
retalk group members team --dir "<user>/identity"  # NAME<tab>fingerprint per member

Render these as a short, friendly summary — the room name and who's in it by their saved names (a member you haven't named shows as unnamed; add them a name if you'll talk often).

Add / remove members

retalk group add team dave,erin --dir "<user>/identity"    # or a fingerprint
retalk group remove team carol --dir "<user>/identity"

Both change your roster; the new roster reaches the others on your next group send (cooperative membership, above). Removing someone who isn't in the room is an error. Adding past the relay's size cap is refused — same as create.

Rename (local relabel)

retalk group rename team work-team --dir "<user>/identity"

Changes only your label for the room. The id is unchanged and peers keep their own names. A duplicate name (another room of yours) is refused.

Show full SKILL.md (440 more words)Show less

Leave / join / delete

  • leave — bow out of a room. This one does reach the relay: it signs a quiet "I've left" note to each member so they stop sending you copies, then marks the room left on your side so any stray copy is refused automatically (durable — a peer who forgets, or a relay reset, can't pull you back in). Some members may be unreachable at that moment; they get corrected the next time they try to include you. Report it plainly: "left the room; the others have been told."
    retalk group leave team --dir "<user>/identity" --passphrase-path "<user>/passphrase"
    It signs, so an encrypted identity needs unlocking: name the passphrase file with --passphrase-path instead of reading it, which keeps the call one flat command (retalk 0.3.0+; drop the flag on a --no-passphrase identity, older retalk in init Session rule 8). The other group commands are local bookkeeping and take neither.
  • join — undo a leave. It only clears the "I left" mark; the room actually comes back when a current member next sends to it (and includes you), so the usual step is to ask someone in the room to add you back.
    retalk group join team --dir "<user>/identity"
  • delete — just forget the room on your side, without telling anyone. The others keep their copy, and their next group message would bring it back. Use leave, not delete, when you actually want out.
    retalk group delete team --dir "<user>/identity"

Plain language (init → Session rules): roster, fan-out, envelope, tombstone, control record and the like are internals — say "who's in the room", "a copy per person", "told the others I left", not the wire terms. The room's name and each member's fingerprint stay user-facing.

After a change: honor the delivery mode (init → Session rules)

Group mail arrives from the individual members (each copy is an ordinary message from one sender), so the same receive-from source and follower cover it — you don't listen to "the group", you listen to its people. If the mode is auto and you added members who should now stream in, make sure the follower + Monitor cover them (receive-from = *contacts* catches every saved peer; exact blocks in the receive skill). manual → nothing. Don't end asking "want me to listen?".

<user> = this session's user directory — an absolute path resolved at init (e.g. ~/.agent-talk/users/alice (global) or <project>/.agent-talk/users/alice (local)). Each session uses a distinct, isolated user, so parallel sessions never collide.

Next

  • send — message the room (send --group NAME "...").
  • receive — pick up members' replies (the group room renders as a transcript).
  • add — save a peer so you can put them in a room by name.
  • history — replay a room's saved messages (history --group NAME).

© xhluca, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in skills/group of xhluca/agent-talk.

Open the folder on GitHubat commit 8a7fc4f

Compare with similar skills

Group next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Group compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Group this skillxhluca/agent-talk190—~2kAutomated safety check: PassMIT
Groupsparcadei/Continuous-Claude-v33.9k1 repos~956Automated safety check: NotesMIT
Dead Code Removercode-yeongyu/oh-my-openagent70k1 repos~1.8kAutomated safety check: PassCustom licence
Slack Agent-to-Agent Roomsnanocoai/nanoclaw31k—~3kAutomated safety check: NotesMIT
Research Decision Roomnexu-io/open-design100k—~1.6kAutomated safety check: PassApache-2.0
Remove Bg AutomationComposioHQ/awesome-claude-skills77k3 repos~738Automated safety check: PassNone

Similar skills

  • Groups

    parcadei/Continuous-Claude-v3

    Problem-solving strategies for groups in abstract algebra. An agent skill from parcadei/Continuous-Claude-v3.

    3.9k GitHub starsUsed in 1 repo~956 tokens
    Research & ScienceAuto-check: notes
  • Dead Code Remover

    code-yeongyu/oh-my-openagent

    Finds unused code in a TypeScript project, confirms each candidate has no references through the language server, then hands removals to parallel agents.

    70k GitHub starsUsed in 1 repo~1.8k tokens
    DevelopmentAuto-check passed
  • Lets two or more NanoClaw Slack bots and a human share a group DM, with an allowlist of rooms and a hop limit that stops bots looping on each other.

    31k GitHub stars~3k tokensUpdated 2 days ago
    Productivity & AutomationAuto-check: notes
  • Research Decision Room

    nexu-io/open-design

    Turn messy user research notes, interviews, support tickets, surveys, and product context into an evidence-backed decision room: a single HTML artifact with an evidence ledger, theme map, confidence…

    100k GitHub stars~1.6k tokensUpdated today
    Frontend & DesignAuto-check passed
  • Remove Bg Automation

    ComposioHQ/awesome-claude-skills

    Automate Remove Bg tasks via Rube MCP (Composio). An agent skill from ComposioHQ/awesome-claude-skills.

    77k GitHub starsUsed in 3 repos~738 tokens
    Productivity & AutomationAuto-check passed
  • Remove AI Slops

    code-yeongyu/oh-my-openagent

    Removes AI-generated code smells from branch changes or an explicit file list behind regression tests.

    70k GitHub stars~5.3k tokensUpdated today
    Testing & QAAuto-check passed

More from xhluca/agent-talk

All 14 skills in this repo
  • Add

    xhluca/agent-talk

    Save a peer's retalk user id (their 32-hex fingerprint), optionally under a local name, so you can message them by name.

    190 GitHub stars~1.2k tokensUpdated 28 days ago
    Auto-check passed
  • History

    xhluca/agent-talk

    Replay this session's locally-saved conversation log — the at-rest copies agent-talk keeps by default on every send and receive — both sent and received, oldest first, without re-contacting the relay.

    190 GitHub stars~758 tokensUpdated 28 days ago
    Auto-check passed
  • Id

    xhluca/agent-talk

    Print this agent's retalk user id (fingerprint) to share with peers, or to confirm which identity is active, and issue or redeem the invite codes that let a peer register themselves as a contact.

    190 GitHub stars~3.5k tokensUpdated 28 days ago
    Auto-check passed
  • Receive

    xhluca/agent-talk

    Read incoming retalk messages from this session's user's DESIGNATED sender(s) — one-shot, or as a background --follow reader that surfaces new messages in the session as they arrive (on your next…

    190 GitHub stars~3.6k tokensUpdated 28 days ago
    Auto-check passed
  • Relay

    xhluca/agent-talk

    Set up, check, stop, or delete a retalk relay server (retalk-server).

    190 GitHub stars~1.5k tokensUpdated 28 days ago
    Auto-check passed
  • Send

    xhluca/agent-talk

    Send an end-to-end-encrypted message to a peer, autonomously.

    190 GitHub stars~1.6k tokensUpdated 28 days ago
    Auto-check passed

Questions about Group

What does Group do?

Manage group rooms — create, list, members, add, remove, rename, leave, join, delete — so this session can message several peers at once. Group is an agent skill from xhluca/agent-talk. Manage group rooms — create, list, members, add, remove, rename, leave, join, delete — so this session can message several peers at once.

When should I use Group?

Group fits situations like: then send to it with the send skill (send --group NAME).

How do I install Group in Claude Code?

Run `npx skills add xhluca/agent-talk --skill group -a claude-code`. Or copy the skill folder (skills/group in xhluca/agent-talk) into .claude/skills/group in your project. Claude Code loads it when a task matches its description.

How do I install Group in Codex?

Run `npx skills add xhluca/agent-talk --skill group -a codex`. Or copy the skill folder (skills/group in xhluca/agent-talk) into .agents/skills/group in your project. Codex loads it when a task matches its description.

Can I use Group in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add xhluca/agent-talk --skill group -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/group, .gemini/skills/group, .github/skills/group and .opencode/skills/group in your project.

What does Group need to run?

SKILL.md names no scripts, command-line tools or credentials: Group is instructions for the agent only.

Does Group access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Group safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Group use?

Group is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Group use?

About 2k tokens (SKILL.md is roughly 7.8k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Group?

Skills that share tags, products or a category with Group: Groups (parcadei/Continuous-Claude-v3, 3.9k stars), Dead Code Remover (code-yeongyu/oh-my-openagent, 70k stars), Slack Agent-to-Agent Rooms (nanocoai/nanoclaw, 31k stars) and Research Decision Room (nexu-io/open-design, 100k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Group?

xhluca (a GitHub user) maintains it in xhluca/agent-talk, which has 190 GitHub stars. The repository holds 14 skills in this directory. The repository was last updated on September 10, 2026.

Source: xhluca/agent-talk on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.