AI tool orchestration

Based on this meeting transcript (an engineering design discussion,
technical review, or feature kickoff), produce a prompt for [Claude
Code | Cursor | v0 | other code-gen tool] that will generate the
code/component described.

The prompt should be self-contained — the downstream tool will not
have access to the transcript. Include all necessary context.

Constraints:
- Specific about language, framework, existing patterns
- States acceptance criteria explicitly
- Notes constraints, style conventions, and anti-patterns
- Provides examples if the meeting referenced patterns to follow
- Single, focused scope (don't bundle multiple changes)

Format:

# Prompt for {Target Tool}
**Target tool:** {Claude Code | Cursor | v0 | etc.}
**Source meeting:** {Meeting name — date}

---

## The task
One to two sentences. The specific code change or component.

## Context
**Codebase / project:**
- Language/framework: {e.g., TypeScript, Next.js 14, App Router}
- Existing patterns to follow: {point to specific files or describe
  conventions}
- Style: {code style guide or example reference}

**The problem being solved:**
Two to three sentences from the meeting. What's broken or missing?
What should be true after this change?

## Specific requirements
1. {Requirement from the meeting}
2. {Requirement}

## Edge cases to handle
- {Edge case mentioned}

## What's out of scope
Things discussed but not part of this task.
- {Item}

## Acceptance criteria
- [ ] {Specific testable criterion}
- [ ] {Specific testable criterion}

## Style / convention notes
- {Convention}: example or pointer

## Anti-patterns / what NOT to do
- {Anti-pattern} — why to avoid

## File-level guidance (if multi-file)
- **Create:** files to create
- **Modify:** files to modify with brief description
- **Don't touch:** files explicitly out of scope

## Examples to reference
{If the meeting referenced existing code patterns, paste them or
point to them}

---

End of prompt.

Based on this meeting transcript (a strategic discussion, customer
feedback session, data review, or problem-framing conversation),
produce a prompt for [Claude | ChatGPT | Perplexity | Notebook LM]
that will produce the analysis or research output discussed.

Format:

# Prompt for {Target Tool}
**Target tool:** {tool name}
**Source meeting:** {Meeting name — date}
**Purpose:** {What I want to do with the output}

---

## The question / goal
Stated clearly. One sentence preferred, two at most.

## Context
Two to four sentences. Why this analysis matters. What decision or
work it will inform. Constraints the analyst should be aware of.

## What I'm trying to understand
Specific sub-questions:
1. {Sub-question}
2. {Sub-question}

## Data / sources to use
{For Perplexity: types of sources to prioritize}
{For Notebook LM: which uploaded sources are relevant}
{For Claude/ChatGPT: data pasted below or attached}

**Data:**
{Paste relevant data, or describe what data the tool has access to,
or attach files}

## Reasoning approach
- {Comparative analysis between X and Y}
- {Pattern identification across N items}
- {Cause/effect tracing}
- {Trend analysis over time}

## Output format
- **Length:** target — brief, moderate, comprehensive
- **Structure:** sections, bullets, table, comparative matrix
- **Audience:** who reads this
- **Tone:** analytical-neutral, executive-summary, internal-team

## What I want to leave out
Specific things NOT to include. Helps focus the output.
- {Topic out of scope}

## Quality criteria
The analysis is good if:
- {Specific criterion}

## Citations / evidence requirements
{For Perplexity: cite sources, prefer X over Y, note recency}
{For Notebook LM: quote source documents}
{For others: when to be confident vs. when to flag uncertainty}

## What to do if you don't have enough information
How the tool should handle gaps — ask clarifying questions, make
clearly-flagged assumptions, return partial analysis with explicit
gaps.

---

End of prompt.

Based on this meeting transcript (a discussion of a document or
communication that needs to be produced), produce a prompt for
[Claude | ChatGPT | other writing AI] that will draft the document.

Format:

# Prompt for {Target Tool}
**Target tool:** {tool name}
**Source meeting:** {Meeting name — date}
**Document type:** {PRD | Email | Memo | Blog post | Report}

---

## What I'm creating
One to two sentences describing the document.

## Audience
- **Who reads this:** specific audience
- **What they already know:** baseline context
- **What they care about:** their priorities
- **What they're skeptical of:** pushbacks to pre-empt

## Goal of the document
What action or understanding should the reader come away with?

## Structure
The document should have these sections:
1. {Section name} — what goes here, one sentence
2. {Section name} — what goes here

Or, if freer-form, the narrative arc:
{Opening → middle → close, as discussed}

## Voice and tone
- {Tone descriptor — analytical / warm / firm}
- **Avoid:** specific tone or language to stay away from

## Source material
The substance to draw on:
{Paste relevant content from the meeting transcript, or quotes, or
supporting documents}

## Length
- **Target:** word count or page count
- **Hard limit:** absolute max

## Specific requirements
- Must include: things the document must cover
- Must mention: people, products, references to name explicitly
- Must NOT include: topics out of scope

## Style constraints
- Brand voice if applicable
- Formatting conventions
- Use of headings, bullets, tables
- No emojis (unless brand voice allows)

## Examples to follow / avoid
{Positive reference if you have one; negative reference if useful}

## Output format
- **Format:** Markdown | Plain text | Specific structure
- **Sign-off / closing:** if applicable
- **Placeholders:** items I'll fill in manually

---

End of prompt.

Based on this meeting transcript (a discussion of a recurring task or
domain you want AI assistance with), produce a system prompt or
project-context document for [Claude Project | Custom GPT | persistent
assistant]. This prompt establishes the context the assistant uses
across every future conversation.

Format:

# System Prompt for {Target Assistant}
**Target:** {Claude Project / Custom GPT / other}
**Purpose:** {What this assistant exists to help with}
**Source:** {Meeting(s)}

---

## Role
You are {role description — be specific. "An assistant" is too generic.
"A product manager for X team helping me think through technical
trade-offs" is useful}.

## Context about me / my work
Two to four paragraphs that establish what the assistant should know:
- **What I work on:** specific products, projects, areas
- **My role:** function, responsibilities, scope
- **Who I work with:** team, stakeholders
- **What I'm currently focused on:** active projects, near-term goals
- **My typical use cases for you:** what I'll typically ask for

## Domain context
- **Product / system specifics:** what the system is, how it works
- **Industry context:** if relevant
- **Key concepts and terminology:** internal language the assistant
  should know and use
- **People I'll reference:** team members, customers, partners

## What I'm trying to accomplish (broadly)
Strategic goals or recurring objectives.

## How I want you to behave
- **Tone:** how you should sound — direct, exploratory, careful
- **Bias toward:** behaviors I want — pushback on weak ideas, ask
  clarifying questions, propose alternatives
- **Avoid:** behaviors I don't want — excessive hedging, generic
  advice, "as an AI" disclaimers
- **When uncertain:** what to do — say so, propose, ask
- **Format preferences:** length, structure, formatting

## Knowledge I'm giving you
Reference documents, specifications, or context the assistant should
treat as authoritative for this project.

## What you DON'T need to know
Topics or contexts the assistant should not assume — to avoid drift.

## How I'll typically engage with you
Pattern of interaction — quick questions, deep dives, iterative
drafting.

---

End of system prompt.