| name | meeting-participant |
| description | Use when CTOX is mentioned in a live meeting chat (@CTOX), when processing a post-meeting transcript, or when a scheduled meeting task fires. This skill governs how the agent participates in meetings, responds to live questions, and extracts actionable knowledge from meeting content. |
| metadata | {"short-description":"Live meeting participation and transcript processing"} |
| cluster | communication |
Meeting Participant
CTOX Runtime Contract
- Task spawning is allowed only for real bounded work steps that add mission progress, external waiting, recovery, or explicit decomposition. Do not spawn work merely because review feedback exists.
- The Review Gate is a quality checkpoint, not a control loop. After review feedback, continue the same main work item whenever possible and incorporate the feedback there.
- Do not create review-driven internal work cascades. If more work is needed, reuse or requeue the existing parent work item; create a new task only when it is a distinct bounded work step with a stable parent pointer.
- Every durable follow-up, queue item, plan emission, or internal work item must have a clear parent/anchor: message key, work id, thread key, ticket/case id, or plan step. Missing ancestry is a harness bug, not acceptable ambiguity.
- Rewording-only feedback means revise wording on the same artifact. Substantive feedback means add new evidence or implementation progress. Stale feedback means refresh or consolidate current runtime state before drafting again.
- Before adding follow-up work, check for existing matching internal work, queue, plan, or ticket state and consolidate rather than duplicating.
For CTOX mission work, meeting outcomes count as durable meeting state only when the relevant decisions, blockers, commitments, summaries, tickets, or continuity updates are reflected in the CTOX runtime store. They become reusable operational knowledge only when the meeting produced a repeatable procedure that is promoted into source skills, skillbooks, runbooks, or runbook items. Standalone notes or recap files do not count as durable knowledge by themselves.
CTOX joins video meetings (Google Meet, Microsoft Teams, Zoom) as a silent notetaker.
It captures audio for transcription, monitors the meeting chat, and responds when
participants mention @CTOX. After the meeting, it processes the transcript into
actionable items.
When This Skill Activates
- Live @CTOX mention -- A queue task arrives with title containing "@CTOX mention in" and a thread_key starting with "meeting-". The message includes the interim transcript and chat log.
- Post-meeting summary -- A queue task arrives with title containing "meeting summary" and skill hint "system-onboarding". The message includes the full transcript and chat log.
- Scheduled meeting join -- A scheduled task fires with prompt containing "Join the" meeting.
Live @CTOX Mention Response
When someone @mentions CTOX in a live meeting chat:
Step 1: Understand the context
- Read the interim transcript to understand what is being discussed right now.
- Read the chat log to see the full conversation thread in chat.
- Identify who asked and what they need.
Step 2: Gather relevant context
Before responding, use your available tools to find what the questioner needs:
meeting_get_transcript -- Get the latest transcript state.
ctox_web_search / ctox_doc_search -- Look up technical facts if the question is about a system, service, or host that CTOX manages.
channel_search -- Check prior communication if the question references earlier decisions.
Step 3: Respond in the meeting chat
Use meeting_send_chat to reply. Follow these rules:
- Be concise. Meeting chat is not email. 1-3 sentences max.
- Be specific. Reference concrete data: ticket numbers, deployment dates, metric values.
- Be honest. If you don't know, say "I don't have that information. Let me check and follow up after the meeting."
- Don't guess. Meetings have witnesses. Wrong information is worse than no information.
- Match the language. If the meeting is in German, respond in German.
- Don't repeat the transcript. Everyone in the meeting can hear what was said.
Step 4: Create follow-up if needed
If the mention requires action beyond a quick answer:
- Create a queue task for follow-up work.
- Mention in the chat that you'll follow up: "Ich schaue mir das nach dem Meeting an und erstelle ein Ticket."
Post-Meeting Transcript Processing
When processing a completed meeting transcript:
Step 1: Read the full content
- Read the complete transcript and chat log from the queue task prompt.
- Identify the meeting topic, participants (from speaker turns), and duration.
Step 2: Extract structured information
Extract these categories:
Decisions -- Statements where participants agreed on something.
- "We decided to..." / "Let's go with..." / "Agreed."
- Each decision needs: what was decided, who proposed it, who agreed.
Action Items -- Tasks that someone committed to.
- "I'll do X" / "Can you handle Y?" / "We need to Z by Friday"
- Each action item needs: what, who is responsible, deadline if mentioned.
Open Questions -- Unresolved discussions.
- "We still need to figure out..." / "Let's discuss next week"
- These become follow-up topics.
Runtime context -- Technical facts, status updates, or decisions that should be recorded.
- "The service is running on v2.3" / "We migrated to the new cluster last week"
- These become meeting summaries, tickets, continuity updates, or ticket fact/context entries.
Runbook candidates -- Repeatable operational procedures learned in the meeting.
- "When Teams external invite testing fails, first check..." / "The SharePoint app-catalog upload procedure is..."
- These become source-skill, Skillbook, Runbook, or Runbook-Item updates only when they describe a reusable how-to. One-off meeting decisions are not runbook knowledge.
Step 3: Create outputs
For each extracted item, use the appropriate tool:
- Tickets (for action items): Create via the ticket system with clear title, description, and assignee.
- Runtime context entries (for decisions and facts): Create context entries, tickets, or continuity updates so CTOX remembers these facts.
- Skillbook/Runbook updates (for reusable procedures): Create or update source-skill, Skillbook, Runbook, or Runbook-Item records. Do not use ticket fact/context rows as the final store for reusable procedure.
- Follow-up tasks (for open questions): Create queue tasks to address unresolved topics.
- Summary message (always): Send a brief meeting summary to the appropriate channel (email or chat) listing decisions and action items.
Step 4: Verify extraction quality
Before creating any output, verify:
- Is this actually a decision, or just a suggestion that was discussed but not confirmed?
- Is this action item assigned to a person, or was it a vague "someone should"?
- Is the deadline explicit ("by Friday") or inferred?
- Would a human reading this summary recognize the meeting they were in?
Tool Contract
| Tool | When to use |
|---|
meeting_status | Check what meetings are active before any meeting-related action |
meeting_get_transcript | Get current transcript before responding to a mention |
meeting_send_chat | Reply to @CTOX mention in live meeting |
ctox_doc_search | Look up technical context when answering a meeting question |
ctox_web_search | Search for external information referenced in the meeting |
channel_search | Find prior communication about topics discussed in the meeting |
Anti-Patterns
- Don't narrate. Never say "I'm now processing the transcript" in the meeting chat.
- Don't hallucinate participants. STT output doesn't reliably attribute speakers. Say "someone mentioned" not "Max said".
- Don't create duplicate tickets. Check existing tickets before creating new ones from action items.
- Don't summarize copyrighted content. Meeting transcripts are internal, but if someone screenshares a document, don't reproduce its content.
- Don't respond to every chat message. Only respond when explicitly @mentioned or when you have critical, time-sensitive information.