scion-conversation
How to use the scion conversation command for managing conversations. Covers listing, creating, viewing, participant management, and message retrieval. Complements the scion-messaging skill.
- License 许可
- Apache-2.0
- 最近更新
- 2026-11-13
- Stars
- ★ 2.1k
- Forks
- 310
01 前言 Front Matter
- name
- scion-conversation
- description
- How to use the scion conversation command for managing conversations. Covers listing, creating, viewing, participant management, and message retrieval. Complements the scion-messaging skill.
02 资源文件 1
- 参考文档agents/openai.yaml82 B
03 SKILL.md 正文
Scion Conversation
Overview
Use scion conversation to inspect and manage conversation metadata,
participants, and message history. Conversation commands require Hub mode.
This command does not send messages. Use scion message to write to a
conversation; see the scion-messaging skill for sending and reply-routing
guidance.
When to Use
- See which conversations you participate in.
- Create a group conversation for multi-agent coordination.
- Read message history or catch up on recent activity.
- List or add participants, or leave a conversation.
- Retrieve a specific message by ID.
When NOT to use: Do not use scion conversation to send messages. Use
scion message for writing into conversations. Avoid creating a conversation
for a one-off direct message; send it with scion message @<agent-name> instead.
Conversation References
Conversation subcommands accept three reference formats:
conv:<uuid>: Direct conversation ID. This is the most common form and always works for a conversation you can access.@<agent-name>: The direct-message conversation with an agent.#<thread-name>: A named thread conversation.
Bare UUIDs are not accepted. Always include the conv: prefix when addressing
a conversation by ID.
Functional Capabilities
Reading: list, get, messages, catch-up, and get-message let you
discover accessible conversations, inspect their metadata, and read all,
recent, or individual messages. Prefer structured output when another command
or agent will consume the result.
Administration: create, join, leave, participants, and
set-default create coordination spaces and manage their membership and
default agent. A conversation name is the positional argument to create, not
a --title flag. join is not idempotent: adding an existing participant
returns HTTP 409, so check participants first when membership is uncertain.
Run scion conversation --help for the full command reference and
scion conversation <subcommand> --help for authoritative usage, arguments,
and flags for an individual subcommand.
Common Patterns
Reply in the original conversation
Read the inbound envelope's conversation.id, add the conv: prefix, and send
the reply with scion message:
scion message conv:a1b2c3d4-... "Reply in the original conversation"
Do not use scion conversation to send the reply. See the scion-messaging
skill for the complete routing rules.
Create a coordination space
scion conversation create "project-x coordination" --json
Capture the returned ID and share it as conv:<id> with participants.
Catch up and manage membership
Use catch-up to review recent activity without rereading the full history.
Before adding a participant whose membership is uncertain, inspect
participants; call join only when they are absent.
Relationship to scion message
The commands are complementary:
| Goal | Command |
|---|---|
| Send into a conversation | scion message conv:<id> "text" |
| Read conversation history | scion conversation messages conv:<id> |
| Read one message | scion conversation get-message conv:<id> <message-id> |
| Inspect or administer a conversation | scion conversation get, participants, join, or leave |
Use the scion-messaging skill for writing and reply routing, including the
structured-markdown format required for multi-sentence messages. Use this
skill for reading and conversation administration.
Anti-Patterns
- Passing a bare UUID instead of a
conv:<uuid>reference; the command rejects bare UUIDs. - Trying to send through
scion conversation; usescion messageinstead. - Treating
scion conversation joinas idempotent; an existing participant produces HTTP 409. - Creating a conversation for a one-off message; use
scion message @<agent>for a direct message.
04 同仓库更多 Skills
team-creation
Create or extend scion agent team templates from a high-level description of roles and workflow. Use when the user describes a multi-agent team, panel, crew, pipeline, or any scenario requiring coordinated LLM agents with distinct roles. Also use when adding new roles to an existing team, modifying workflows, or restructuring agent coordination. Produces ready-to-use template directories in .scion/templates/ by default, or in a custom path if specified.
scion-web-serving
How to serve a web page, app, API or other HTTP service from inside your agent so a person or program can reach it through the Hub's port proxy: scion expose and unexpose, the URL to hand over, what the proxy carries, and scion tunnel for full apps. The link serves web protocols only; for a non-HTTP port, point the person to scion tunnel. Use before running scion expose. For web pages and apps, also read references/web-apps.md.
scion-scheduler
Schedule one-shot and recurring events for agents using the Scion CLI. Covers when to use scheduled events vs inline message delays, recurring cron schedules, lifecycle management, and the blocked-wait pairing rule.
scion-messaging
Use when sending messages to users or agents with scion message, choosing conversations and mentions, or interpreting incoming messages. Covers user-facing formatting, agent delivery, and reading conversation history.
scion-cli-operations
Operational constraints for Scion agents running in containerized sandboxes. Covers prohibited commands and hub-only API access. Complements the scion CLI reference and messaging skills.
scion-agent-manage
Manage concurrent LLM-based code agents with scion - orchestrate parallel agents with isolated workspaces, troubleshoot and recover stuck agents