Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.productbrain.com/llms.txt

Use this file to discover all available pages before exploring further.

You’re shipping with Claude Code, Cursor, or similar AI agents. They’re productive — but they operate in a vacuum. Each session starts from scratch. The agent doesn’t know what you shipped yesterday, what’s coming next, or how this feature connects to the business.

​
The problem

Without shared state, you’re the bottleneck. You brief the agent every session, manually track what’s done, and hope you don’t duplicate work or miss a dependency. The agent is fast but blind.

​
How ProductBrain helps

​
Shared state

Your planning tree is the shared context between you and your agents. The agent reads the tree to understand:
  • What goal this work serves
  • What approach it’s delivering
  • What jobs remain in the current iteration
  • What “done” looks like for each job

​
Agent workflow

1. Agent reads the tree via API
2. Agent finds the current iteration
3. Agent picks the next undone job
4. Agent reads the job description (what to verify)
   and notes (how to build it)
5. Agent does the work
6. Agent verifies the checkpoint
7. Agent marks the job done via API
8. Repeat

​
You stay strategic

While the agent executes, you work in the Planning Tree. Review approaches, adjust priorities, add new needs based on what you’re learning. The agent picks up changes automatically — next time it reads the tree, it sees the updated plan.

​
Setup

  1. Generate an API key from Settings
  2. Copy the quickstart snippet (includes docs URL, base URL, key, project ID)
  3. Paste into your agent session
  4. The agent reads the LLM Operations Guide and starts working

​
What changes

  • Planning and status survive compaction effortlessly — the tree is external memory that outlasts any session
  • Work is automatically tracked — jobs marked done as the agent ships
  • You can run multiple agents on different approaches simultaneously
  • The Delivery Map shows real-time progress across all your bets
  • Agent output is verifiable — each job describes what to check