Automated Notes & Action

Turn meeting transcripts into structured summaries with auto-created action items in Jira, Linear, or Todoist — all via OpenClaw Ultra.

Core System Overview

This system bridges the gap between "we discussed it" and "it's tracked and assigned." Feed it a meeting transcript (paste, file, or auto-watch folder), and the agent outputs a structured summary plus tracked tasks assigned to the right people.

System Layer	Core Function	Output Result
Ingestion Layer	Transcript input via paste, file drop, or folder watch	Clean parsed text with speaker labels
Analysis Layer	Key decisions, discussion topics, sentiment, speaker attribution	Structured meeting summary
Extraction Layer	Action item identification with owner & deadline detection	Structured task list per attendee
Distribution Layer	Post summary to Slack/Discord, create tickets in Jira/Linear/Todoist	Notified team + tracked tasks
Follow-up Layer	Deadline reminders, overdue escalation, completion tracking	Closed-loop accountability

Prerequisites

Item	Requirement
OpenClaw Ultra	Installed and running
Project Management API (Pick One)	Jira / Linear / Todoist — API token or access
Team Communication (Optional)	Slack or Discord webhook for posting summaries
Transcript Source	Manual paste, Zoom/Google Meet transcript files, or Otter.ai export

Step 0 — Initialize Meeting Notes System

Set up OpenClaw Ultra as your automated meeting secretary.

Operation Steps

1. Open OpenClaw Ultra new chat session
2. Have your project management tool API credentials ready
3. Paste initialization prompt

Ready-to-Use Prompt

Act as my automated meeting notes & action item system.

I use:
- Project management: [Jira / Linear / Todoist]
- Team communication: [Slack / Discord / none]
- Transcript source: [manual paste / file folder / Otter.ai]

Configure the system to:
- accept meeting transcripts in any format
- extract key decisions, discussion topics, and action items
- create tracked tasks with assignees and deadlines
- post summary to team channel
- follow up on overdue items

Step 1 — Manual Transcription Processing

The simplest path: paste a transcript and get structured output instantly.

Prompt

I just finished a meeting. Here's the transcript:

[paste transcript here]

Please:
1. Write a concise summary (max 5 bullet points) covering key decisions and topics
2. Extract ALL action items. For each one identify:
   - What needs to be done
   - Who is responsible (match names to my team)
   - Deadline (if mentioned, otherwise mark "TBD")
3. Create a [Jira/Linear/Todoist] ticket for each action item, assigned to the right person
4. Post the full summary to [#channel] in [Slack/Discord]

INFO

Speaker-labeled transcripts (from Zoom, Google Meet, or Otter.ai) produce better action item attribution. Plain text works too — the agent infers ownership from context.

Step 2 — Standardize Output Format

Ensure every meeting summary follows a consistent structure.

Prompt

When writing meeting summaries, always use this structure:

- **Date & Attendees** at the top
- **Key Decisions** — numbered list of what was decided
- **Action Items** — table with columns: Task, Owner, Deadline, Status
- **Open Questions** — anything unresolved that needs follow-up

Save this as my permanent meeting notes format rule.

Step 3 — Automate with Folder Watch

For recurring meetings, set up automatic transcript processing.

3.1 Folder-Based Auto Pipeline

Prompt

Set up a recurring task: every 30 minutes, check [~/meeting-transcripts/]
for new .txt, .vtt, or .srt files.

When you find one:
1. Parse the transcript into a structured summary with action items
2. Create tasks in [Jira/Linear/Todoist] for each action item
3. Post the summary to [#channel] in [Slack/Discord]
4. Move the processed file to [~/meeting-transcripts/processed/]

For each action item with a deadline, set a reminder to ping the assignee
in Slack one day before it's due.

3.2 Supported Transcript Formats

Prompt

Configure support for these transcript formats:
- .txt — plain text with optional [Speaker] labels
- .vtt — WebVTT from Zoom/Google Meet (includes timestamps)
- .srt — SubRip from any video platform
- Otter.ai export — JSON or plain text

Detect format automatically and parse accordingly.

Step 4 — Connect to Project Management Tools

4.1 Jira Integration

Prompt

Connect to my Jira workspace:
- Domain: [your-domain].atlassian.net
- Project: [project key]
- API Token: [token]

When creating action items, use these conventions:
- Story type for new tasks
- Bug type for issues
- Priority based on deadline proximity
- Label: "meeting" for traceability

4.2 Linear Integration

Prompt

Connect to my Linear workspace:
- API Key: [key]
- Team: [team name]
- Project: [project name]

Create action items as Linear issues with:
- Assignee from transcript name matching
- Label "meeting-notes"
- Due date parsed from transcript

4.3 Todoist Integration

Prompt

Connect to my Todoist:
- API Token: [token]
- Project: [project name]

Create action items as Todoist tasks with:
- Assignee via shared project
- Due date from transcript
- Section: "🟡 In Progress"

Step 5 — Follow-Up & Accountability

Close the loop with deadline reminders and completion tracking.

Prompt

Set up follow-up automation:

Daily at 9 AM:
- Check all action items created from meetings in the last 7 days
- Report which are: completed, in progress, overdue, or no status update

For overdue items:
- Ping the assignee in Slack/Discord with: "Reminder: [task] was due [date]"
- If overdue by 3+ days, escalate to me with a summary

Weekly at 5 PM Friday:
- Send a "meeting notes accountability report" to [#channel]
- Show: total action items created, completed rate, top overdue owners

Final Workflow Logic

Meeting Happens → Transcript Captured → Auto-Ingested →
Structured Summary Generated → Action Items Extracted →
Tickets Created in Jira/Linear/Todoist → Summary Posted to Team →
Deadline Reminders → Overdue Escalation → Weekly Accountability Report

Practical Usage Tips

1. For best results, use Zoom/Google Meet's built-in transcript feature (VTT format) — timestamps help the agent attribute statements accurately
2. Start with manual paste mode (Step 1) before setting up folder watch — validate output quality first
3. Pair with the Todoist Task Manager workflow for full agent-side task visibility
4. Keep a team member name mapping in a memory file so the agent correctly matches names to accounts across tools