# Communication

> Model conversation intent, speech style, turn-taking, listening, repair, and nonverbal cues. Use before or after social interaction, dialogue, negotiation, apology, request, gossip, or conflict.

- **Type:** Skill
- **Install:** `agentstack add skill-alexanderj-carter-agentsociety2-agent-skills-communication`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [AlexanderJ-Carter](https://agentstack.voostack.com/s/alexanderj-carter)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [AlexanderJ-Carter](https://github.com/AlexanderJ-Carter)
- **Source:** https://github.com/AlexanderJ-Carter/AgentSociety2-Agent-Skills/tree/main/skills/communication
- **Website:** http://alexander.xin/AgentSociety2-Agent-Skills/

## Install

```sh
agentstack add skill-alexanderj-carter-agentsociety2-agent-skills-communication
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Communication

## Purpose

Maintain how the agent communicates with others: what it wants to say, how direct it is, whether it listens, how it repairs misunderstandings, and what nonverbal signals matter.

## Internal Logic (One Sentence)

Use participants, relationship, culture, norms, emotion, and conversational context to select communicative intent, style, turn state, and repair strategy, then write `state/communication.json`.

## Use When

Use this skill around conversations, requests, refusals, apologies, negotiation, gossip, teaching, conflict, small talk, or public speaking.

## Procedure

1. Read `state/observation.txt`, `state/observation_ctx.json`, `state/relationships.json`, `state/culture.json`, `state/norms.json`, `state/emotion.json`, and previous `state/communication.json` if present.
2. Identify participants, relationship, setting, power distance, audience, and active goal.
3. Choose a communicative intent: inform, ask, request, refuse, comfort, apologize, persuade, coordinate, joke, signal status, avoid, or repair.
4. Select style: direct/indirect, formal/informal, warm/cold, brief/elaborate, assertive/deferential.
5. Track turn state: whether the agent should speak, listen, clarify, or end the interaction.
6. Write `state/communication.json`.

## Write

Write `state/communication.json`.

## Output Schema

```json
{
  "_meta": {
    "skill": "communication",
    "purpose": "Conversation intent, style, turn state, and repair strategy."
  },
  "_summary": "Use a polite indirect request and listen for Alice's response.",
  "participants": ["alice"],
  "intent": "request",
  "style": {
    "directness": 0.35,
    "formality": 0.62,
    "warmth": 0.74,
    "assertiveness": 0.42
  },
  "turn_state": "speak_next",
  "repair_needed": false,
  "recommended_utterance_goal": "Ask Alice if she can share the schedule without pressuring her."
}
```

## Notes

Communication is not just text generation. It should reflect relationship, emotion, norms, culture, and purpose. If uncertain, prefer asking clarifying questions or listening.

For theory notes, see `references/theory.md`.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [AlexanderJ-Carter](https://github.com/AlexanderJ-Carter)
- **Source:** [AlexanderJ-Carter/AgentSociety2-Agent-Skills](https://github.com/AlexanderJ-Carter/AgentSociety2-Agent-Skills)
- **License:** MIT
- **Homepage:** http://alexander.xin/AgentSociety2-Agent-Skills/

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-alexanderj-carter-agentsociety2-agent-skills-communication
- Seller: https://agentstack.voostack.com/s/alexanderj-carter
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.