PM AI playbook: 17 skills, 20+ integrations, context graph

You know that feeling when you're writing a PRD and need to reference three Jira tickets, two Slack threads, a Figma mock, and last quarter's strategy doc? You end up with 15 browser tabs and a Notion page that's outdated the moment you finish it. PM work spans too many tools—none of which were built for the messy, collaborative, half-creative-half-structured way you actually think. You're a tourist in Figma, a visitor in Jira, and your "product knowledge" lives in scattered docs no one updates.

You clone the repo, open it in Cursor IDE, and type "onboard." The agent asks about your company, role, goals, and tools, then generates personalized configuration files. It creates three layers: rules (always-on guidance about your product context), skills (on-demand capabilities like PRD writing or experiment design), and agents (specialized assistants that run workflows like feedback analysis). The key innovation is the context graph—a memory layer where every agent output gets saved, so your feedback analyzer in March knows what February found, and your exec update pulls from decisions, plans, and reviews automatically.

• 17 PM skills out of the box — PRD writer, experiment designer, launch posts, one-pagers, exec updates, and more work immediately without any configuration
• Context graph memory — agents read from and write to a shared memory, so your feedback analyzer knows what previous analyses found and your exec updates pull from decisions automatically
• 20+ MCP integrations — connect Slack, Jira, Linear, Notion, Figma, Google Drive, Amplitude, Salesforce, Databricks, and more; the setup script generates the config, you just add API keys
• Personalized onboarding — answer questions about your role, goals, and tools; the system generates rules and agents tailored to your specific product context
• Company research agent — give it a company name and it automatically researches LinkedIn, 10-K filings, earnings calls, and job postings to build strategy context
• Knowledge layer with real examples — includes fully-worked examples from Spotify, Netflix, Shopify, and Uber using public investor data to show how to structure strategy docs

Who It Is For

If you're a PM who spends more time context-switching between tools than actually thinking about product strategy, this is for you. Especially valuable for PMs at companies where knowledge is scattered across Slack, Notion, Jira, and Google Docs with no single source of truth. Not useful if you don't use Cursor IDE or if your team has rigid tool requirements that prevent adding new integrations.

Worth Exploring?

Yes, if you already use Cursor or are willing to try it. The project is new (first commit Feb 2026) but actively maintained (last update Mar 17, 2026) with 81 stars and 36 forks in its first month. The 15+ skills work immediately with zero setup, and the MCP integrations are optional—you can start simple and add connections later. The context graph concept is genuinely novel for PM tools. The main caveat: it requires Cursor IDE, so if your team standardizes on VS Code or another editor, you'll be working outside your normal environment.