Meeting Prep

Pulls a focused prep block for an upcoming meeting. Surfaces the last 3 meetings + open action items in both directions. Goes in today's journal as a single block.

How to invoke

• /meeting-prep Bob — prep for next meeting with people/Bob
• /meeting-prep juan-miguel — match the namespaced name
• Natural: "prep for my 1:1 with <person> tomorrow", "what's the history with <person>", "I have a meeting with <person> — what do I need to remember"

Steps

1. Resolve the person

The argument is a person name (or a partial). Resolve to a people/<name> page:

1. Try exact match: people/Bob
2. If not found, fuzzy match against existing people/* pages — show 2-3 options if ambiguous
3. If still no match, ask the user to clarify

Use mcp__mcp-logseq__list_pages to get the existing namespace; filter to entries starting with people/.

2. Get backlinks

Use mcp__mcp-logseq__get_page_backlinks on the resolved person page. Returns every block that references [[people/<name>]] — usually meeting blocks with attendees:: [[people/<name>]].

3. Filter to meeting blocks

Each backlink is a block. Keep only blocks that are meeting captures (typically: link to [[meetings]], contain attendees::, time::, etc.). Sort by date descending (most recent first).

4. Pull open action items

Within those meeting blocks, find any TODO items. Classify:

• They owe me: TODO items where the user's name is NOT in parens (or no parens at all), AND the action is in a meeting they attended
• I owe them: TODO items where the user's first name IS in the parens (e.g. (Alice) if user is named Alice), in a meeting they attended

Skip items already marked DONE or CANCELED.

5. Write the prep block

Append to today's journal:

- ## Prep — {person's name}
  - ### Last 3 meetings
    - {date} — {title}
      - {1-2 line gist from the summary}
    - {date} — {title}
      - {1-2 line gist}
    - {date} — {title}
      - {1-2 line gist}
  - ### They owe me ({count})
    - TODO {item} (from {meeting title}, {date})
  - ### I owe them ({count})
    - TODO {item} (from {meeting title}, {date})
  - ### Suggested agenda
    - 
    - 

Skip empty sections. Leave "Suggested agenda" blank — the user fills it in.

6. Note voice — notes, not docs

The meeting gists are written in your words — keep them note voice: terse, personal, reader is future-you.

• State the point; no hand-holding phrases ("for the user", "in order to")
• No markdown emphasis (**bold**, *italic*)
• Gist over 12 words? Try cutting half
• Plain language over code jargon — drop class/method names unless the gist is meaningless without them; keep numeric thresholds
• Swaps: "implemented" → "added", "leveraged"/"utilized" → "used", "applies to" → "needs"

(Full version: logseq-note skill, "Notes are notes, not docs".)

Worked example

User: /meeting-prep alice

Output:

- ## Prep — Alice
  - ### Last 3 meetings
    - Jun 17, 2026 — 1:1 with Alice
      - Discussed Q3 priorities, agreed to scope down a feature
    - Jun 10, 2026 — Team standup
      - Alice flagged a project as on-track
    - Jun 3, 2026 — Tech showcase demo
      - Alice showed a new feature, customer success liked it
  - ### They owe me (1)
    - TODO Send me the Q3 plan (from 1:1, Jun 17)
  - ### I owe them (2)
    - TODO Review Alice's PR (from standup, Jun 10)
    - TODO Decide on launch date by Friday (from 1:1, Jun 17)
  - ### Suggested agenda
    - 
    - 

Hard constraints

• One block in today's journal, not a separate page.
• Last 3 meetings only. More is noise — this is prep, not history.
• Distinguish ownership of TODOs. Failing to do this is the main reason prep blocks become useless.
• Quote the source meeting for every action item (so user can click through to context).
• No invented action items. Only items literally present in past meeting blocks.

Voice

No tool-call narration. No decorative tables or emoji. No long raw error-log dumps unless user asks. Standard tech acronyms OK (DB/API/MCP); never invent new abbreviations.

Abbreviate prose words (auth/config/req/res/fn/impl) — prose words only, never real code symbols or function names. Strip conjunctions; use arrows for causality (X → Y). One word when one word enough. Code symbols, function names, API names, error strings: never abbreviate.

No filler. No hedging. No "you", "your", "we". Keep sentences intact. Professional but tight.

Example

❌ "Sure! I can help you prepare for your meeting with Alice. Let me look up her page in your people namespace and find your last three meetings together."

✅ "Prep block written for Alice: 3 past meetings, 1 open action item they owe, 2 owed to them. Suggested agenda left blank."

Don't

• Don't include meetings older than 6 months unless the user asked for full history
• Don't write a long narrative summary — bullets only
• Don't auto-fill the "Suggested agenda" — that's the user's prep work
• Don't include DONE/CANCELED action items
• Don't tag the prep block — it's in the journal, that's enough

Edge cases

• Person not found: ask user to clarify, suggest 2-3 closest matches
• No past meetings: say so. "No previous meetings found with people/Bob. Starting fresh — what's the agenda?"
• Group meetings (3+ attendees): prep block focuses on what this person contributed. Don't repeat unrelated discussion.
• Multiple people with same firstname: use the full namespaced name (people/juan-miguel vs people/juan)
• Person attended but didn't speak / no notes about them: skip their meeting in the "Last 3" list, find an earlier one
• Old prep blocks accumulate. Prep blocks aren't tagged, so they don't clutter hubs. But after months they add visual noise in journal pages. The user can collapse old prep blocks in Logseq or occasionally archive/delete them.