Meeting prep briefer

Meeting Prep Briefer

This skill generates a structured per-meeting briefing document for busy managers and individual contributors. Given a daily schedule (pasted inline or read from a local file) and any available workspace context files (project notes, participant notes, previous meeting notes), it produces a brief for each meeting with participants, context, open questions, and a suggested agenda — entirely from local files, no external integrations required.

Input: Daily schedule (inline text or local file path) + workspace context files (auto-scanned).

Output: Structured markdown brief (inline response). Optionally saved to output/meeting-brief-YYYY-MM-DD.md on request.

---

Language Detection

Detect the user's language from their message:

• If Russian (or contains Cyrillic): respond in Russian
• If English (or other Latin-script language): respond in English
• If ambiguous: default to English

---

Instructions

Step 1: Parse Input

1. Check if the user provided an inline schedule (list of meetings in their message) or a file path.

   • Inline: parse meetings directly from the message.
   • File path: read the specified file from the workspace. If the file does not exist, return: "File not found: {path}. Please check the path or paste the schedule directly."

2. If no schedule is provided at all, return: "Please paste your meeting schedule or provide a path to a schedule file (e.g., today.md)." Stop — do not generate output.

3. Parse the schedule into a list of meetings. Each meeting must have at minimum a time and a title. Participants are optional.

   • Accepted formats: 10:00 — Team Sync, 10:00 Team Sync (Alice, Bob), 10:00 | Team Sync | Alice, Bob, or similar time + title patterns.
   • If the schedule is completely unparseable (no recognisable time+title structure), return: "Could not parse meeting schedule. Please use a format like: 10:00 — Team Sync (Alice, Bob)." Stop.

4. Determine the brief date: use today's date unless the user specifies another date.

---

Step 2: Scan Workspace for Context

1. For each parsed meeting, search the workspace for relevant context files:

   • Match files by: meeting title keywords, participant names, or project names appearing in the schedule.
   • Look in: root workspace, notes/, output/, input/, context/ folders.
   • Accepted file types: .md, .txt.

2. For each matched file, extract:

   • Relevant headings and their content (limit to 150 words per file to keep briefs concise).
   • Any open action items or questions found in the file.

3. If no matching files are found for a meeting: note "No context files found" for that meeting — do not skip the meeting.

---

Step 3: Assemble Briefs

For each meeting, build a section with four subsections:

Participants

• List names and roles if provided in the schedule.
• If no participants listed: note "No participants listed in schedule."

Context

• List 2–4 bullet points extracted from matched workspace files.
• Each bullet: From \filename`: [key point]`.
• If no context files matched: "No context files found — add relevant notes to your workspace."

Open Questions

• Extract any open questions or action items from matched files.
• If none found: leave as an empty prompt line *(none listed — add questions before the meeting)*.

Suggested Agenda

• Generate 2–4 agenda items based on the meeting title, participants, and context.
• Keep each item to one short phrase with a suggested time allocation.

---

Step 4: Output Brief

1. Assemble all meeting sections into a single markdown document:

   • Header: # Meeting Brief — YYYY-MM-DD
   • Each meeting: ## HH:MM — Meeting Title followed by four subsections
   • Meetings separated by ---

2. Display inline in the chat.

3. If the user says "save" or "save to file": write the brief to output/meeting-brief-YYYY-MM-DD.md and confirm the path.

---

Edge Cases

• No context files for a specific meeting: Produce the brief anyway; fill Context with placeholder message.
• Schedule has time + title only (no participants): Skip Participants section; add flag: "No participants listed — add names to schedule for richer brief."
• Multiple files match one meeting's keywords: Include all matched files as context sources; list each filename.
• User provides a past or future date: Use that date for the brief header and output filename.
• Very long meeting title (>60 chars): Truncate display title in the header to 60 chars; use full title in the section heading.

---

Negative Cases

• No schedule provided: Return usage message and stop. Do not generate partial output.
• File path not found: Return file-not-found message with the exact path. Do not guess alternate paths.
• Schedule unparseable: Return format hint and stop. Do not attempt to guess meeting times.

---

Output Format

# Meeting Brief — 2026-04-28

---

## 10:00 — Team Sync

**Participants:** Alice (PM), Bob (Eng), Carol (Design)

**Context:**
- From `project-notes.md`: Sprint 12 ends Friday; 3 open stories remain.
- From `alice-notes.md`: Concerns raised about API timeline dependency.

**Open Questions:**
- Is the API dependency resolved before Friday?
- Who owns the design handoff?

**Suggested Agenda:**
1. Sprint status check (5 min)
2. API dependency decision (10 min)
3. Next steps & owners (5 min)

---

## 14:00 — 1-on-1 with Bob

**Participants:** Bob (Eng)

**Context:**
- No context files found — add relevant notes to your workspace.

**Open Questions:**
*(none listed — add questions before the meeting)*

**Suggested Agenda:**
1. Check in on current blockers (10 min)
2. Career development update (10 min)