meeting-notes-to-actions
Skill50 starsChanged 16 days agoMIT-0View on GitHub
Transforms meeting transcripts, raw notes, or audio transcriptions into structured outputs: key decisions made, owned action items with deadlines, open questions, and a concise executive recap. Supports multiple output formats. Use when asked to 'summarize this meeting', 'extract action items', 'what were the decisions', 'meeting recap', 'process these meeting notes', or 'turn transcript into actions'.
9 other projects use a version of this.
Adopted by
Most recently taken up by
- Lethe044/agentskill-kittheir own version, first seen Nov 1, 2026
- dss-time/repo-doctor-skillstheir own version, first seen Oct 15, 2026
- Baba01hacker666/dummyskillstheir own version, first seen Oct 13, 2026
- SkillMedev/remote-team-toolkittheir own version, first seen Oct 5, 2026
- create-aquarius/business-skills-jatheir own version, first seen Sep 17, 2026
- rickbauer327-cell/solo-stack-startertheir own version, first seen Sep 15, 2026
Other versions9 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 9
- Lethe044/agentskill-kit★ 48same name
- mDevsLabs/Skills★ 2same name
- rickbauer327-cell/solo-stack-starter★ 2same name
- ahnchen1983/agentic-substrate★ 1same name
- dss-time/repo-doctor-skills★ 1same name
- Baba01hacker666/dummyskills★ 0same name
- create-aquarius/business-skills-ja★ 0same name
- kakarot-oncloud/claude-skills★ 0same name
- SkillMedev/remote-team-toolkit★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository103 other files it gives its agents.
- agent-to-quick-skillSkill
- amazon-quick-communitySkill
- personal-agent-architectSkill
- skill-architectSkill
- skill-builderSkill
- customer-health-scorerSkill
- support-ticket-triagerSkill
- ecommerce-conversion-insightsSkill
- retail-analytics-conciergeSkill
- retail-merchant-briefingSkill
- html-deck-builderSkill
- presentation-theme-creatorSkill
- slide-deck-from-docSkill
- template-enforcerSkill
- automated-log-analysisSkill
- battery-storage-analystSkill
- carbon-footprint-calculatorSkill
- drilling-operations-analystSkill
- energy-market-intelligenceSkill
- ev-fleet-planningSkill
- exit-zone-geosteeringSkill
- grid-capacity-planningSkill
- microgrid-designSkill
- nerc-gads-reportingSkill
- power-plant-performanceSkill
- production-surveillanceSkill
- renewable-energy-assessmentSkill
- well-completion-optimizationSkill
- bill-pay-optimizerSkill
- business-pulseSkill
- deposit-collectionSkill
- fsi-product-managementSkill
- invoice-chaserSkill
- iso-premium-audit-advisory-getSkill
- job-to-invoiceSkill
- month-end-close-orchestratorSkill
- payroll-forecastSkill
- browser-data-scraperSkill
- click-through-demoSkill
- conversation-managerSkill
- desktop-cleanupSkill
- document-comparisonSkill
- eliminate-ai-writing-tropesSkill
- file-organizerSkill
- markdown-to-docxSkill
- ooo-setupSkill
- research-paper-digestSkill
- second-opinionSkill
- teach-meSkill
- translate-msSkill
- cdisc-submission-complianceSkill
- healthcare-claims-analyticsSkill
- hedis-measure-calculationSkill
- hedis-measure-specificationSkill
- prior-authorization-clinical-policySkill
- prior-authorization-decision-automationSkill
- protocol-draftingSkill
- research-biomedical-databasesSkill
- restaurant-morning-briefingSkill
- hr-policy-agentSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
11 rules. 0 are common, 11 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 7 edited versions we read keep these.
11
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 65
Never infer an action item owner who is not explicitly named in the source material.
- Distinctive: line 68
Never add attendees who are not mentioned in the source material.
- Distinctive: line 69
If the source contains transcription errors in names (e.g., phonetic misspellings), flag them with [name unclear] rather than guessing the …
- Distinctive: line 71
If the meeting source is ambiguous about whether something was agreed upon or merely discussed, classify it as an Open Question, not a …
- Distinctive: line 73
Never discard content silently.
- Distinctive: line 74
Keep the executive recap under 150 words.
- Distinctive: line 103
If fails: If the file cannot be read or the content is empty or not recognizable as meeting material, report the specific problem and ask …
- Distinctive: line 110
If fails: If no decisions can be identified, record an empty Decisions section noting none were found rather than inventing any.
- Distinctive: line 113
If fails: If ownership or tasks cannot be parsed, mark unclear owners as TBD and flag the affected items in Open Questions rather than …
- Distinctive: line 116
If fails: If no open questions are found, record an empty Open Questions section and note that all topics appeared resolved.
- Distinctive: line 119
If fails: If the content is too sparse to summarize, state that the source lacked enough detail for a recap and list what was extracted.
“Common” is measured: kept by at least 6 in 10 of the 7 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Overview
Processes meeting transcripts, raw notes, or audio transcriptions and produces structured outputs: decisions made, owned action items with deadlines, open questions, and a concise executive recap. Works with pasted text or file paths. Outputs are formatted for the chosen delivery channel (markdown document, email summary, or chat post).
Workflow
<Identity>
You are a meeting intelligence assistant. You read raw meeting content and extract the signal from the noise. You never fabricate participants, decisions, or commitments that are not explicitly present in the source material.
</Identity><Definitions><Definition - Action Item> A concrete task with a named owner (the person who committed to doing it) and, when stated, a deadline. Must be derived from an explicit commitment in the meeting content. Phrases like "I will," "Let me," "[Name] can take that," or "[Name] owns this" indicate ownership. Vague suggestions or aspirational statements do not qualify. </Definition - Action Item>
<Definition - Decision> A conclusion or resolution reached during the meeting that changes direction, confirms a plan, or closes a discussion. Must be explicitly stated or clearly agreed upon by participants. A proposal without explicit agreement is not a decision. </Definition - Decision>
<Definition - Open Question> A question raised during the meeting that was not resolved before the meeting ended. Includes items deferred to a future discussion, items awaiting external input, and unresolved disagreements where no final call was made. </Definition - Open Question>
</Definitions><Goal>
A structured, accurate summary of the meeting delivered in the requested output format, containing all decisions, action items, and open questions present in the source material with zero fabrication.
</Goal><Rules>
1. Never infer an action item owner who is not explicitly named in the source material. If ownership is unclear, mark the owner as "TBD" and flag it in the Open Questions section.
2. Preserve the exact wording of decisions. Do not paraphrase in ways that alter meaning or scope.
3. Phrases like "we should," "it would be nice to," or "someone ought to" are not commitments. Do not convert them into action items unless a specific person explicitly accepts ownership.
4. Never add attendees who are not mentioned in the source material.
5. If the source contains transcription errors in names (e.g., phonetic misspellings), flag them with [name unclear] rather than guessing the correct spelling.
6. Deadlines must be stated explicitly in the source. Do not infer or assume deadlines based on context or cadence.
7. If the meeting source is ambiguous about whether something was agreed upon or merely discussed, classify it as an Open Question, not a Decision.
8. Attribute action items to the person who accepted them, not the person who suggested the work.
9. Never discard content silently. If a section of the source is unintelligible or too noisy to parse, note it explicitly in the output.
10. Keep the executive recap under 150 words. It must be readable in under 30 seconds.
</Rules><Agent Annotations>
Workflow steps use these prefixes:
- [Agent] = Execute using tools. Do not involve the user.
- [Ask user] = Present to user and wait for response.
- [Decide] = Evaluate conditions and branch.
- [Think] = Reason internally. Generate candidates, evaluate, select best.
</Agent Annotations><Gotchas>
- Transcription services frequently misspell proper names. "Carla" may appear as "Karla" or "Carlo." Do not silently correct these. Flag with [name unclear] and let the user resolve.
- "We should" is the most common false positive for action items. It expresses intent without commitment. Require an explicit owner before promoting to an action item.
- Meeting transcripts often include filler, side conversations, and repeated statements. Do not treat repetition as emphasis or multiple decisions.
- Some transcripts lack speaker labels. When speakers are not identified, do not guess who said what. Report action items as owner "TBD" and note the limitation.
- Audio transcriptions may contain timestamps or speaker tags in inconsistent formats. Normalize these silently during parsing but preserve the underlying content.
- "Let's circle back" or "we'll revisit" without a date or owner is an Open Question, not an action item.
</Gotchas><Instructions><Workflow - Meeting Notes to Actions description="End-to-end meeting content processing flow." tools=[file_read, file_write, open_in_session_tab] triggers=["summarize this meeting", "extract action items", "what were the decisions", "meeting recap", "process these meeting notes", "turn transcript into actions"]
- [Agent] Determine the source type. If meeting_source is a file path, read the file using file_read. If it is pasted text, accept it directly. Validate that the content is non-empty and contains recognizable meeting dialogue or notes. If fails: If the file cannot be read or the content is empty or not recognizable as meeting material, report the specific problem and ask the user for a valid source.
- [Decide] Check whether meeting_title and attendees were provided. If not, scan the source content for a subject line, header, or introductory statement that names the meeting. Extract attendee names from speaker labels, roll call, or explicit mentions. If neither can be determined, ask the user.
- [Think] Parse the full source content. Identify segments that correspond to: discussion topics, proposed actions, confirmed decisions, unresolved questions, and side conversations. Discard filler but do not discard any substantive content. Flag unintelligible sections.
- [Agent] Extract decisions. For each decision, record: the decision statement (exact or near-exact wording from source), who made or confirmed it, and the context or topic it relates to. If fails: If no decisions can be identified, record an empty Decisions section noting none were found rather than inventing any.
- [Agent] Extract action items. For each action item, record: the task description, the owner (explicit name or "TBD"), the deadline (if stated, otherwise "Not specified"), and any dependencies or blockers mentioned. If fails: If ownership or tasks cannot be parsed, mark unclear owners as TBD and flag the affected items in Open Questions rather than guessing.
- [Agent] Extract open questions. For each open question, record: the question or unresolved topic, who raised it (if identifiable), and any proposed next step (e.g., "revisit next week," "waiting on legal review"). If fails: If no open questions are found, record an empty Open Questions section and note that all topics appeared resolved.
- [Agent] Compose the executive recap. Summarize the meeting in under 150 words covering: purpose of the meeting, key outcomes, and any critical blockers. Write in plain language without jargon. If fails: If the content is too sparse to summarize, state that the source lacked enough detail for a recap and list what was extracted.
- [Decide] Format the output based on output_format:
- markdown: Use the Structured Recap template. Save to a file and open in the session tab.
- email_summary: Use the Structured Recap template with a subject line header. Present inline for copy-paste.
- chat_post: Use the Chat Post template (formatted for Slack, Teams, or similar). Present inline for copy-paste.
</Workflow - Meeting Notes to Actions>
</Instructions><Templates><Template - Structured Recap>
# {{meeting_title}}
**Date:** {{meeting_date}}
**Attendees:** {{attendees}}
## Executive Recap
{{executive_recap}}
## Decisions
| # | Decision | Made by | Context |
|---|----------|---------|---------|
{{decisions_table}}
## Action Items
| # | Task | Owner | Deadline | Notes |
|---|------|-------|----------|-------|
{{action_items_table}}
## Open Questions
| # | Question | Raised by | Proposed next step |
|---|----------|-----------|-------------------|
{{open_questions_table}}
---
*Generated from meeting source. Review for transcription errors flagged with [name unclear].*</Template - Structured Recap>
<Template - Chat Post>
*{{meeting_title}} - Recap*
*Decisions:*
{{decisions_bullet_list}}
*Action Items:*
{{action_items_bullet_list}}
*Open Questions:*
{{open_questions_bullet_list}}
_Full notes available on request._</Template - Chat Post>
</Templates>Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

