Most of these tools are read-only: they return instructions enriched with what earlier stages produced, and your model does the work. That makes the sequencing enforced rather than suggested — asking for buyer research before market intelligence returns a blocked message naming what to run first. Whether that helps depends entirely on whether you want to follow this particular playbook.
An MCP server that turns a go-to-market playbook into a sequence of tools. Each one checks that its prerequisites are done, loads what the earlier stages produced, and returns the instructions for the current step; three separate tools handle all state and file writes.
- Entry points for different appetites: scout for the full pipeline, rapid_test for a cheap signal test, tournament to evaluate several ideas at once, portfolio_triage to rank what you already have, passive_deploy for marketplace assets
- Research and offer: autonomy, market_intel, research, build_blocks, stress_test, unit_economics — with the economics gate before any asset is built
- Build and launch: name_lock, platform, product, deploy, qa, validate_prep
- Validation: validate_check for the daily read, validate_decide for the end-of-window verdict, feedback to route fixes
- Traffic: traffic_strategy, channels, creative_test, funnel_optimize, scale, traffic_analytics, dream_100
- Organic growth: content_engine, content_repurpose, seo_check
- Ongoing: status, daily_check, lessons, voice_extract, bold_action, revenue_review, rapid_check, rapid_graduate, rapid_status, passive_check, passive_compound, passive_portfolio
- State and output: update_pipeline_state, save_asset and capture_learning are the only tools that write anything
Node. A project directory holds the pipeline state, the learnings file and the generated assets; it is set through your client config or falls back to the working directory. Some of the packaged quality checks are Python scripts.
One command — npx -y asset-factory-mcp