gtm

GTM Strategy Generator

A generic, codebase-driven GTM skill. Reads a repository, interviews the user across multiple sessions, and produces a complete set of strategy documents.

Output

4 Markdown files in {repo}/docs/gtm/ plus a state file:

docs/gtm/
├── gtm-state.yaml            # Session state (skill internal use)
├── 01-brand-strategy.md       # Who we are
├── 02-market-landscape.md     # The battlefield
├── 03-messaging-framework.md  # How we talk
├── 04-channel-playbook.md     # Where we show up

Modes

Mode	Trigger	Description
Initial	No gtm-state.yaml exists	Deep research: codebase scan → multi-session interviews → draft → review
Update	gtm-state.yaml exists with status: complete	Weekly: scan git diff → confirm with user → update affected sections
Resume	gtm-state.yaml exists with status != complete	Continue from where the last session left off

---

State Management

On first run, create docs/gtm/gtm-state.yaml:

project: {repo name}
repo_path: {absolute path}
created: {YYYY-MM-DD}
last_updated: {YYYY-MM-DD}
last_scan_commit: {commit hash}

status: scanning  # scanning | interviewing | drafting | review | complete

phases:
  scan:
    status: pending      # pending | in_progress | complete
    depth: null           # quick | standard | deep
    completed_at: null
  interview:
    status: pending
    current_topic: null   # product | audience | competition | brand | channels
    completed_topics: []
    pending_topics:
      - product
      - audience
      - competition
      - brand
      - channels
  draft:
    status: pending
    completed_at: null
  review:
    status: pending
    completed_at: null

# Accumulated findings from interviews (append after each session)
findings:
  product: {}
  audience: {}
  competition: {}
  brand: {}
  channels: {}

At the start of every session:

1. Read gtm-state.yaml
2. Display a brief summary: "Last session we covered [X]. Conclusions: [Y]. Today we continue with [Z]."
3. Ask user: continue, go back and revise a topic, or skip to a specific topic.