Growth Scout

Scout opportunities before you write code.

Open-source, agent-agnostic skill package for product growth scouting. Use it in your own repos — copy, fork, or clone.

Growth Scout is a portable, agent-agnostic skill (markdown, not an app) that helps coding agents recommend what feature to build next with the highest chance of helping a product grow.

Works with Cursor, Claude Code, Codex, GitHub Copilot, OpenCode, Aider, and any agent that can read project instructions.

Core question

What should we build next that has the highest chance of helping this product grow?

Install

Growth Scout is a Markdown skill package, so there is no package manager install step. Add the files to the product repo you want an agent to analyze.

Option A: Copy into a product repo

git clone https://github.com/008993368-yaz/Growth-Scout.git
cd your-product-repo
cp /path/to/Growth-Scout/SKILL.md .
cp /path/to/Growth-Scout/AGENTS.md .
cp /path/to/Growth-Scout/VERSION .
cp -R /path/to/Growth-Scout/references .
cp -R /path/to/Growth-Scout/templates .
cp -R /path/to/Growth-Scout/examples .

Then prompt your agent: Follow Growth Scout (SKILL.md) and recommend what to build next.

For Claude Code, Codex, GitHub Copilot, OpenCode, Aider, and other setups, see integrations/README.md.

Option B: Bootstrap install

From the Growth Scout package root, install skill files and generate a draft product map in one step:

git clone https://github.com/008993368-yaz/Growth-Scout.git
node Growth-Scout/scripts/install-growth-scout.mjs /path/to/your-product-repo

Optional layouts: --layout root (default, copies to repo root) or --layout cursor (.cursor/skills/growth-scout/). Use --dry-run to preview or --force to overwrite an existing install. The script prints a ready-to-use first agent prompt when done.

Quick start (any agent)

1. Copy SKILL.md, references/, templates/, examples/, and VERSION into your product project (or clone this repo).
2. Prompt: Follow Growth Scout (SKILL.md) and recommend what to build next.
3. Optional per-agent wiring: integrations/README.md.

Example outputs (fictional Chartflow narrative)

What it combines

Product context, codebase awareness, competitor gaps, market signals (evidence-rated), growth scoring, feasibility, and PRD templates.

Example prompts

Follow Growth Scout (SKILL.md): what should we build next?
Run a Growth Scout product scan. Label unknowns clearly.
Turn the top opportunity into a PRD and GitHub issues.

Package layout

AGENTS.md
SKILL.md
integrations/     # Optional per-agent setup (Cursor is one option)
mcp/              # Optional MCP server (competitor + work-item signals)
references/
templates/
examples/
.cursor/rules/    # Optional Cursor-only hook

Tooling

Zero-dependency Node scripts (no npm install required):

# Install skill files + first-scan draft into a product repo
node scripts/install-growth-scout.mjs /path/to/product-repo
node scripts/install-growth-scout.mjs /path/to/product-repo --layout cursor
node scripts/install-growth-scout.mjs --help

# Pre-fill a product-map draft from a target repo (stdout or file)
node scripts/repo-inventory.mjs .
node scripts/repo-inventory.mjs /path/to/product --out product-map-draft.md
node scripts/repo-inventory.mjs --help

# Validate markdown links in core docs (also runs in CI)
node scripts/validate-links.mjs

Optional Growth Scout MCP (any MCP-capable agent): build and connect the stdio server per integrations/mcp.md. Server package: mcp/growth-scout-mcp/.

cd mcp/growth-scout-mcp && npm install && npm test

Auto-detected inventory fields are labeled inferred; unknowns stay blank for the agent or user to fill.

Versioning

The VERSION file tracks the skill package revision (currently 0.6.1). Copy it into product repos alongside SKILL.md so teams know which Growth Scout revision they pinned. See CHANGELOG.md and CONTRIBUTING.md.

vs generic PM skills

Codebase-aware scoring, evidence labels (known/inferred/unknown), competitor gap buckets, and explicit research limits when browsing is unavailable.

Safety

Do not invent market demand. Cap scores when evidence is weak. Use references/validation-experiments.md when proof is missing.

Roadmap

Optional MCP for competitor fetch, industry reference packs, GitHub release tags aligned with VERSION.

Tagline: Scout opportunities before you write code.

License

MIT — use, modify, and distribute freely with attribution.