Meeting Notes to Review-Ready Records — Actions, Stakeholders, Incident Timelines

Cómo usar esta colección

# From messy notes to something you can hand to a human This collection is a short progression, not a topic keyword list. Every member is a plain-text prompt you paste into an ordinary AI chat that accepts text — no install, no account connection, no API key. Nothing here takes photographs, sends messages, schedules work, or updates a system; all four are drafting aids whose output a person still has to review. ## Who this is for Anyone who leaves a meeting or an incident with raw notes and has to turn them into something a colleague or manager can act on: team leads, project coordinators, support and ops staff, event organisers, and small teams without a dedicated note-taker. ## Prerequisites - An ordinary AI chat that accepts pasted text. - Your own raw material: minutes, a transcript excerpt, a chat log, an email thread, or an incident log. - The right to paste that material into that tool. Meeting notes and logs often contain names, commercial plans, private remarks, or customer data. Redact what you are not allowed to share, and follow your own organisation's rules on third-party AI tools. ## The progression, and why in this order **Step 1 — Stakeholder Interest Map from Meeting Notes** (`<https://tokrepo.com/assets/50a4b899-c4cc-48fb-91a2-1cfe30aa46f5>`). Build the neutral picture first. This prompt reads only your pasted notes and produces a table of every named party's stated position, the basis given, open questions and an evidence-strength label (`explicit`, `implied`, `unclear`), plus shared ground, live disagreements, floating issues and a verification list. Do this before writing actions, because knowing who actually said what keeps you from assigning work to the wrong person. **Step 2 — Meeting Minutes to Action Register Prompt** (`<https://tokrepo.com/assets/a7cdc10d-46a7-4273-a857-651747981c9b>`). Now extract the work. It converts a rough minute draft into a meeting snapshot, a Markdown action register (number, faithful wording, owner, due date, status, source quote, notes), a decisions list, open questions and a confidence note. Its central rule is that a blank beats a guess: where your minutes never name an owner or a date, those columns stay empty and section D raises the question instead. It also keeps tentative wording tentative, so "maybe by Friday" is not upgraded into a deadline. **Step 3 — Incident Log to Factual Timeline and Draft** (`<https://tokrepo.com/assets/f0f33817-8fde-4806-a732-ab3e4fe69cd2>`). Use this when the thing you are documenting is not a meeting but a sequence of events — an outage, a service complaint, a project slip. It orders entries only by the timestamps given, shows conflicting times side by side instead of picking one, drops untimed entries into their own group rather than guessing a position, and drafts a follow-up message that you review and send yourself. **Step 4 — Event Schedule to Photo Shot List Prompt** (`<https://tokrepo.com/assets/a909b218-97cc-4a4c-bab7-11a9106fa9aa>`). The planning counterpart. It turns a supplied event schedule into a checkable shot list — one image per moment, framing and orientation, the single must-show element, a rationale and a status — with coverage gaps and a review checklist. It is included because it demonstrates the same discipline applied forward instead of backward: don't invent a person, a role, a time or a location; mark the row needs more detail instead. ## How to check any output before you rely on it - Every row or table entry traces to a phrase in the material you pasted. Delete anything that does not. - Owner, due date and status stay blank wherever your source was silent. - Tentative words ("maybe", "should", "could") stay tentative. - Conflicting times or positions are shown side by side, not resolved. - Nothing is labelled decided, agreed, or majority unless your notes say so. - Open questions list at least one question per field left blank. ## Honest limitations - Output quality depends entirely on how complete your notes are. A fragment produces a partial register or timeline, labelled as such. - Evidence-strength labels such as `implied` or `unclear` mean the model is inferring; treat them as prompts to verify with real people, not as facts. - No member can confirm that a described action actually happened, verify a time against an external clock, or browse, message anyone, or update any system. - None of these prompts gives legal, HR, medical or financial advice, and none of them is a final record of commitments. - The source materials state source reviewed; runtime not tested. The supplied evidence contains no executed run, so behaviour across specific chat products is not claimed, and the worked examples inside the prompts are labelled fictional output-shape illustrations, not test results. **Fictional illustration (not a real event, and not a test result):** a coordinator pastes stand-up notes naming Priya, Tom and an unnamed "someone". After step 1 the map shows Priya and Tom with explicit positions and the unnamed task as a floating issue. After step 2 the register keeps the owner blank on that task and section D asks who owns it. Nothing is invented and nothing is sent.