- Home
- Skills
- Prompt & Skill Engineering
- Template Skillmd Converter
Works with the AI Agents you already use
Template Skillmd Converter
Turns a working prompt or template into a SKILL.md an agent will actually load: a name, trigger phrases, ordered steps, the prompt's rules as validation checks, and a scope
$5
/moSubscription · cancel anytime
Secure checkout via Stripe
- Always the latest version
- Delivered live to your agent via MCP
- Cancel anytime, access ends at period end
Template Skillmd Converter
Example session with this skill installed
Turn this prompt into a skill
Act as a meeting-notes writer. I will paste a transcript of a [MEETING_TYPE] meeting. Write: 1) a three-line summary, 2) decisions made, 3) action items as a table with owner and due date, 4) open questions. Only use what is in the transcript; if an owner or date is missing write TBD. Keep it under 300 words. Transcript: [TRANSCRIPT]
- Read your context and instructions
- Compiled the template skillmd converter
Here's your prompt converted into a skill. The two slots, [MEETING_TYPE] and [TRANSCRIPT], are no longer placeholders. The user just pastes the transcript and can mention the kind of meeting in the same message. The skill uses the meeting type if one is given, works it out from the transcript if not, and otherwise falls back to "Meeting". Your rules (use only the transcript, TBD for a missing owner or date, under 300 words) are now checks in the Validation section.
---
name: meeting-transcript-to-notes
description: Turns a pasted meeting transcript into short notes - a three-line summary, decisions made, an action-item table with owner and due date, and open questions - built only from what the transcript says. Activate when a team member, manager or project lead pastes a transcript and says "write up notes from this meeting", "summarize this standup transcript", or "pull the action items out of this call", optionally naming the kind of meeting (standup, client call, retro, planning, board meeting).
---
# Meeting Transcript to Notes
## Purpose
Meeting transcripts are long, and the parts people actually need - what was decided, who owns what by when, what is still unresolved - get buried in them. Hand-written notes fix that but drift: an owner gets guessed, a date gets invented, a half-agreed idea gets recorded as a decision. This skill turns the transcript the user pastes into a fixed four-part set of notes under 300 words, using only what was said. Anything the transcript does not state - an owner, a due date - is written as TBD instead of being filled in.
## When to Use
1. The user pastes a meeting transcript and asks for notes, minutes or a write-up.
2. The user wants the action items from a call, with who owns each one and when it is due.
3. A standup, retro, planning session or client call needs a short summary to send round afterwards.
4. The user wants to know what was actually decided in a meeting, separated from what was only discussed.
5. An auto-generated transcript (Zoom, Teams, Meet, Otter) needs turning into something a colleague can read in a minute.
6. The user names the kind of meeting along with the transcript ("this was our weekly design review") and wants notes framed for it.
7. A long discussion left questions unresolved and the user wants them listed so they can go on the next agenda.
### SCOPE — This skill does NOT handle:
- Meetings with no transcript or recording text supplied - it does not reconstruct a meeting from memory or a description.
- Transcribing audio or video; the input must already be text.
- Full verbatim minutes, legal or board-compliance minutes, or anything that must exceed 300 words.
- Assigning owners or deadlines the meeting did not set, or recommending what the team should do next.
- Sending, posting or filing the notes anywhere.
## Process
1. **Confirm the Transcript** — Locate the transcript text in the user's message; if none is present, or it is only a description of a meeting, stop and ask the user to paste the transcript rather than writing notes from nothing. Output: the transcript to work from.
2. **Set the Meeting Type** — Take the kind of meeting from the user's message (for example "client call" or "sprint retro"); if they did not say, use what the transcript makes plain, and otherwise call it simply "Meeting". Output: the label used in the notes heading.
3. **Write the Summary** — Write exactly three lines covering what the meeting was for, what it mainly covered, and where it ended up. Output: the three-line summary.
4. **List the Decisions** — Record only points the participants clearly agreed or settled, one bullet each; ideas floated without agreement go to open questions, not here. If nothing was decided, write "None recorded in the transcript." Output: the decisions list.
5. **Build the Action-Item Table** — Make a table with the columns Action, Owner and Due, one row per task someone took on. Use the owner and date exactly as stated; where either is missing or unclear, write TBD. If there are no action items, write "None recorded in the transcript." in place of the table. Output: the action-item table.
6. **List the Open Questions** — Record questions raised and not answered, and topics deferred to later, one bullet each; if there are none, write "None recorded in the transcript." Output: the open-questions list.
7. **Assemble and Trim** — Render the notes under the heading "<Meeting type> notes" with four sections in this order: Summary, Decisions, Action Items, Open Questions. Count the words; if the total is 300 or more, shorten wording in the summary, decisions and questions first, and keep every action-item row. Output: the finished notes, under 300 words.
## Validation
- [ ] The notes have four sections in this order: Summary, Decisions, Action Items, Open Questions.
- [ ] The Summary is exactly three lines.
- [ ] Action items are a table with Action, Owner and Due columns, and every owner or date the transcript does not state reads TBD.
- [ ] Every decision, action item, owner, date and open question traces to something said in the transcript; nothing comes from outside it.
- [ ] Each empty section reads "None recorded in the transcript." rather than being left out.
- [ ] The finished notes are under 300 words.
What I added beyond the original prompt:
No transcript: if the user only describes a meeting or forgets to paste the text, the skill asks for the transcript instead of making notes up.
Empty sections: a section with nothing in it reads "None recorded in the transcript." instead of disappearing, so the four-part shape never changes.
Decisions vs. ideas: only points people clearly agreed count as decisions. Ideas raised without agreement go under open questions.
Cutting to 300 words: the summary, decisions and questions get shortened first, and no action-item row is ever removed.
Scope limits: it won't transcribe audio, write full or compliance minutes, assign owners the meeting didn't set, or send the notes anywhere. These limits keep it from firing on the wrong requests.
Report
- Name:
meeting-transcript-to-notes - Size: about 4,800 characters, roughly 1,150 tokens
Install path: ~/.claude/skills/meeting-transcript-to-notes/SKILL.md (or .claude/skills/meeting-transcript-to-notes/SKILL.md inside a project to scope it to that repo)
Subscription · always the latest version
- Secure checkout via Stripe
- Cancel anytime
- Delivered live via MCP
About this skill
For prompts that work but live in a notes file, so you paste them by hand every time. It converts a prompt or template into a deployable SKILL.md with the parts that make an agent pick it up on its own and behave the same way each time.
What it does
It states what the prompt produces and for whom, then writes the frontmatter: a kebab-case name, and a description with three or more quoted trigger phrases. The prompt's slots become inputs the user simply mentions in their message, so there's no fill-in-the-blank section. Its logic becomes ordered process steps with one output each. Its hidden rules (length limits, formats, "only use what's in the transcript") become validation checks, and a scope block says what the skill won't do, so it doesn't fire on the wrong requests. It ends with the file's size, a token estimate and the install path.
Try it first
Paste a prompt you reuse and say "turn this prompt into a skill". Check that the description would match the way you'd actually ask for it.
What it needs
Nothing to install: no scripts, no network access and no API keys. Write access if you want the file saved.
It converts one prompt into one skill; it doesn't write new prompts from scratch or benchmark them. Skills that need bundled scripts or assets are out of scope; it writes the SKILL.md only. It keeps the prompt's intent, so a prompt with a flaw becomes a skill with the same flaw, which you can then fix with a skill auditor. Trigger phrases are inferred from the prompt's purpose, so add your own if your users ask differently.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 27 days ago