meeting-facilitation
Facilitates structured multi-agent meetings for collaborative decision-making and problem resolution. Use when: running agent meetings, coordinating multi-agent discussions, or facilitating collaborative problem-solving sessions.
- License 许可
- AGPL-3.0
- 最近更新
- 2026-11-14
- Stars
- ★ 25
- Forks
- 9
01 前言 Front Matter
- name
- meeting-facilitation
- owner
- pm
- scope
- common
- status
- active
- version
- 1.5.0
- metadata
- { "type": "process", "triggers": [ "meeting", "agent discussion", "collaborative decision", "multi-agent coordination", "facilitate meeting" ] }
- description
- Facilitates structured multi-agent meetings for collaborative decision-making and problem resolution. Use when: running agent meetings, coordinating multi-agent discussions, or facilitating collaborative problem-solving sessions.
- last_reviewed
- 1794268800000
- prerequisites
- []
03 SKILL.md 正文
Context
This skill defines the multi-agent meeting facilitation flow. It is self-contained: the
skill body below is the specification the PM follows when orchestrating a meeting. The
legacy /meeting slash command was retired on 2026-09-26 — invoke this skill directly
by name.
When to Use
Invoke this skill when the user requests:
- A structured multi-agent discussion on a named topic
- Facilitating collaborative decision-making across specialist agents
- Coordinating agent discussions for design reviews, problem-solving, or planning
Execution Steps
The skill runs the facilitation flow directly:
- Agenda setting: state the topic, objectives, participants, and round count. The round count is a meeting parameter fixed here — never reduced silently mid-meeting.
- Opening round (round 1): each participating agent states its position and supporting evidence on the topic.
- Facilitator agenda pass: after round 1, the facilitator (or PM) distills the contributions into 2-5 named agenda issues reflecting where positions spread (criteria and recommendations), and states the issue list before round 2.
- Deliberation rounds (round 2+): run the remaining rounds issue by issue. For
each agenda issue, every participating agent MUST:
- address at least two named peers with an explicit agree / rebut / ask;
- answer every question another participant addressed to it in the previous round;
- state an explicit stance line:
changed: <from> → <to>orno change: <position>; - stay on the agenda issue — new topics go to the facilitator for the issue list. The PM facilitates and keeps contributions on-agenda.
- Outcome synthesis: a cross-domain agent (or the PM) synthesizes agreements, open points, and a proposed recommendation per issue.
- Transcript logging: write the transcript to
memory/meeting-YYYY-MM-DD-[slug].md.
Output Format
Meeting transcript written to memory/meeting-YYYY-MM-DD-[slug].md containing, in order:
- Agenda, objectives, participants, and the issue list from the facilitator agenda pass
- Discussion by issue: attributed exchanges (who said what to whom), round by round
- Agreements reached per issue
- Preserved dissent: recorded disagreements transcribed VERBATIM — never summarized, never averaged into consensus (Governance Rule 3)
- Open questions carried forward
- Action items with owner assignments
- Proposed recommendation per issue, explicitly labelled a PROPOSAL for the user's approval — never a decision or PASS stamp (Governance Rule 2)
Governance Rules
All meetings facilitated through this skill MUST uphold three invariants:
- Dissent seat: at least one participant is designated as a red-team / dissenting role whose duty is to challenge the emerging consensus.
- PROPOSAL, never decision: the synthesized outcome is a proposal for the user's approval — the meeting itself does not decide.
- Dissent preserved verbatim: recorded disagreements are transcribed as stated — never summarized away or averaged into consensus.
Related Skills
project-review— uses meeting-facilitation for Gemini CLI parallel dispatchteam-builder— may invoke meetings during team assembly (Phase 0)
04 同仓库更多 Skills
i18n-layout
Text layout and encoding guidance: character encoding (UTF-8, legacy Korean code pages, BOM hazards), line endings, RTL/bidi handling, and script-specific font selection (Hangul focus). Use when: encoding is corrupted or mojibake appears, an RTL locale must render correctly, fonts must be chosen for a script, or CRLF/BOM issues surface.
i18n-layout
Text layout and encoding guidance: character encoding (UTF-8, legacy Korean code pages, BOM hazards), line endings, RTL/bidi handling, and script-specific font selection (Hangul focus). Use when: encoding is corrupted or mojibake appears, an RTL locale must render correctly, fonts must be chosen for a script, or CRLF/BOM issues surface.
i18n-layout
Text layout and encoding guidance: character encoding (UTF-8, legacy Korean code pages, BOM hazards), line endings, RTL/bidi handling, and script-specific font selection (Hangul focus). Use when: encoding is corrupted or mojibake appears, an RTL locale must render correctly, fonts must be chosen for a script, or CRLF/BOM issues surface.
i18n-layout
Text layout and encoding guidance: character encoding (UTF-8, legacy Korean code pages, BOM hazards), line endings, RTL/bidi handling, and script-specific font selection (Hangul focus). Use when: encoding is corrupted or mojibake appears, an RTL locale must render correctly, fonts must be chosen for a script, or CRLF/BOM issues surface.
i18n-layout
Text layout and encoding guidance: character encoding (UTF-8, legacy Korean code pages, BOM hazards), line endings, RTL/bidi handling, and script-specific font selection (Hangul focus). Use when: encoding is corrupted or mojibake appears, an RTL locale must render correctly, fonts must be chosen for a script, or CRLF/BOM issues surface.
i18n-layout
Text layout and encoding guidance: character encoding (UTF-8, legacy Korean code pages, BOM hazards), line endings, RTL/bidi handling, and script-specific font selection (Hangul focus). Use when: encoding is corrupted or mojibake appears, an RTL locale must render correctly, fonts must be chosen for a script, or CRLF/BOM issues surface.