Esta página se muestra en inglés. Una traducción al español está en curso.
分析Oct 7, 2026·6 min de lectura

Volunteer Debrief Lessons-Learned Register Prompt

Turn volunteer debrief transcripts into a sourced lessons-learned table, separating observations from participant suggestions.

Listo para agents

Instalación lista para agent

Este activo puede instalarse después de elegir el runtime, revisar el plan y ejecutar el comando correspondiente.

Native · 96/100Política: permitir
Superficie agent
Cualquier agent MCP/CLI
Tipo
Prompt
Instalación
Single
Confianza
Confianza: Established
Entrada
PROMPT.md
Comando de instalación directa
npx -y tokrepo@latest install 12c084d7-c312-4ecd-9011-3a5155fefbf5 --target codex

Ejecutar después de confirmar el plan con dry-run.

Start here

Paste your raw volunteer event debrief transcript where the prompt says to. The prompt is plain text, so an ordinary AI chat that accepts text is enough. No terminal or API setup is needed.

What this prompt does

It converts a messy debrief transcript into a Markdown table called "Lessons-Learned Register" with eight columns: Item ID, Topic, What Happened, Source, Suggested Change, Proposed By, Confidence, and Needs Human Follow-Up. It also adds a "Gaps and Ambiguities" section.

The core idea: observations and suggestions stay in separate columns and are never merged. Only changes a participant actually proposed are recorded; otherwise the cell reads "None stated." The prompt includes a fictional worked example and a pass/fail check for that example.

How to use it

  1. Copy the full prompt text (appended separately).
  2. Paste it into an ordinary AI chat that accepts text.
  3. Paste your transcript below the prompt, as the prompt instructs.
  4. Send it and review the resulting table.

How to check the output

  • Every "What Happened" cell traces to a named speaker or identifiable comment in your transcript.
  • No "Suggested Change" appears without a matching "Proposed By."
  • No recommendations appear in the assistant's own voice.
  • If the transcript is too short or disorganized to yield at least three distinct items, the prompt should say so and ask you for more context.

Inputs, permissions and limits

  • Input: one debrief transcript containing participant names or roles where possible. Missing names limit the Source and Proposed By columns.
  • Permission: you are the one pasting the transcript, so you decide what to share. Transcripts may include personal names; remove anything you should not share before pasting.
  • Limits: the prompt writes only a draft register for a human organizer to review. It does not browse, send, or schedule anything, and it does not give legal, safety, or financial advice. Such topics are logged for human follow-up.

Source reviewed; runtime not tested.

FAQ

What if my transcript has no speaker names? The Source and Proposed By columns may end up thin. The prompt asks you for additional context when it cannot extract enough distinct items.

Does it invent recommendations? No. The rules forbid inferring a change that was not spoken or written; unstated changes are marked "None stated."

Complete reusable prompt

You are helping me turn a raw volunteer event debrief transcript into a lessons-learned register. I will paste the transcript below. Your job is to extract and organize the content that is actually present in the transcript — nothing more.

WHAT TO PRODUCE

Create a Markdown table called "Lessons-Learned Register" with these columns:

  1. Item ID (L1, L2, L3...)
  2. Topic (short label, e.g. check-in, supplies, parking)
  3. What Happened (a factual statement grounded in the transcript — who observed what, where, and when if stated)
  4. Source (the speaker's name or role as given in the transcript, plus a short quote fragment under 15 words)
  5. Suggested Change (only if a participant actually proposed one; write "None stated" if not)
  6. Proposed By (name/role of the person who suggested it; "—" if none)
  7. Confidence (High / Medium / Low — based only on how clearly the transcript supports the 'What Happened' entry)
  8. Needs Human Follow-Up (Yes/No, plus a one-line reason)

RULES FOR SEPARATING FACTS FROM SUGGESTIONS

  • A "What Happened" entry must describe something that occurred during or around the event. Do not rephrase a suggestion as an observation.
  • A "Suggested Change" entry must come from an identifiable participant in the transcript. Never invent, merge, or infer a recommendation that was not spoken or written.
  • If a participant mixes observation and opinion in one sentence, split them into separate columns but keep the same Item ID.
  • If the transcript contains vague statements ("it was fine," "people seemed happy"), record them as Low confidence and note in the follow-up column that specifics are missing.
  • If a topic is raised but no clear outcome or change is stated, still log it with Suggested Change = "None stated" and Needs Human Follow-Up = Yes.

UNCERTAINTY AND MISSING INPUT HANDLING

  • After the table, add a section "Gaps and Ambiguities" listing every place where the transcript was unclear, contradictory, or silent on an important point.
  • List any speaker whose role or name could not be determined from the transcript.
  • If the transcript is too short or too disorganized to extract at least three distinct items, say so and ask me what additional context I can supply.

REVIEW CHECKS BEFORE YOU FINISH

  1. Verify every "What Happened" cell can be traced to at least one named source or identifiable comment in the transcript.
  2. Verify no "Suggested Change" appears without a matching "Proposed By."
  3. Verify no two Item IDs describe the same event from different angles unless you explicitly note the overlap in the follow-up column.
  4. Confirm the register contains no recommendations written in your own voice.

BOUNDARIES

  • Do not browse, send, or schedule anything. You are only producing a draft register for a human organizer to review.
  • Do not give advice on legal, safety, or financial matters. If such topics arise, log them as items needing human follow-up.
  • Do not add filler rows to reach a target count.

WORKED EXAMPLE (fictional, for shape only)

Input fragment: "Priya: The check-in table only had one person until 9:30, so the line got long. Tom: We should have two people from the start next time. Priya: Also the name badges ran out by 10."

Expected output row: | L1 | Check-in staffing | Only one volunteer at the check-in table until 9:30, causing a long line | Priya, organizer ("line got long") | Have two people at check-in from the start | Tom | High | Yes — confirm actual arrival numbers |

A pass/fail check on this example: the row passes if "What Happened" contains only the staffing observation and the long line, while "Suggested Change" contains only Tom's proposal. The row fails if the suggested change is merged into the observation column, or if Tom's name is missing from Proposed By.

Now paste your transcript below. I will review the register and fill in any follow-up items myself.

References and reuse

Original TokRepo prompt · CC BY 4.0. Reference documents retain their own rights.

🙏

Fuente y agradecimientos

Original TokRepo prompt, CC BY 4.0. Reference: ChatGPT release notes. External reference material retains its own rights.

Discusión

Inicia sesión para unirte a la discusión.
Aún no hay comentarios. Sé el primero en compartir tus ideas.

Activos relacionados