Scattered notes collapse under pressure
Decisions, task state, and open risks drift across chats, markdown files, temporary scratchpads, and half-remembered plans.
Early product, publicly explained
Structured project state for AI-native builders.
TreeMA gives scattered project context a durable home. Keep structure, decisions, risks, and AI-generated proposals explicit enough to inspect, review, and trust without turning your project into another pile of notes.
- Canonical core
- `.treema/project_state.json` and staged analysis artifacts
- Human control
- AI proposes updates, humans approve canonical change
- Product shape
- Browser app, Electron desktop shell, and CLI helpers
Problem
AI moves fast when context stays coherent.
Project meaning becomes implicit
Builders know what matters right now, but the logic behind scope, tradeoffs, and priorities rarely survives in a durable structure.
AI context turns fragile
Without a stable workspace contract, every new prompt has to reconstruct the same mental model from incomplete evidence.
How TreeMA Works
One explicit workspace, three durable loops.
TreeMA keeps project state explicit first, then layers analysis and AI review on top of that contract instead of replacing it.
Record canonical state
Structure the project map, current tasks, decisions, and risks in a workspace that stays legible outside any single chat or session.
Scan the real project
Analyze a real repository into staged artifacts that separate observed facts, inference, uncertainty, and execution readiness.
Review AI proposals with context
Let AI suggest updates, but keep canonical mutation review-driven so state stays explicit and durable.
Core Capabilities
Built around the current MVP, not future promises.
Canonical workspace
Keep project structure, task state, and context rooted in a deterministic `.treema` workspace contract.
Project Scan
Generate staged inventory, analysis, architecture, and execution artifacts from a real project folder with validator-gated outputs.
Human-reviewed AI updates
Review proposal diffs before any canonical change so AI remains an assistant, not the source of truth.
Durable project history
Preserve decisions, risk movement, and narrow focus updates in a workspace that survives across sessions and tooling changes.
Screens
The product already has real surfaces.
These placeholders reserve the slots for product captures based on the existing app views. Replace them with real exports without changing layout or asset paths.
Why Explicit State Matters
Explicit structure beats invisible context debt.
What stays visible
- Current project shape, not just current conversation
- Risks and decisions separated from tasks
- Observed facts separated from inference and uncertainty
- Execution readiness gated by validator status
TreeMA is not trying to be a full PM suite. It exists to keep project state inspectable enough that humans and AI can operate on the same durable frame.
Product Status
An early product with a narrow, working contract.
TreeMA is in MVP shape. The current surface is intentionally focused on local workflows, explicit workspace files, and reviewable AI assistance.
Available now
- Local app server and Electron shell
- Workspace init, load, snapshot, and export flows
- Project Scan and Quick Scan analysis surfaces
- Read-only AI proposal review
Deliberately not here yet
- Autonomous writes to canonical state
- Real-time collaboration and team permissions
- Workflow automation beyond the narrow MVP loop
- PM-suite breadth like chat, estimation, or sprint tooling
Run Locally
Open the existing product, not a mock.
TreeMA already runs as a local browser app and Electron desktop shell. Start with the browser flow, then move into desktop if you want the same UI with native folder access and IPC-backed actions.
Browser app
npm run serveThen open http://localhost:4173.
Desktop shell
npm install
npm run desktopThe Electron shell loads the same app surface from the local repository.
Workspace CLI
npm run treema -- init /path/to/project "My Project"
npm run treema -- analyze /path/to/projectInitialize a `.treema` workspace and scan a real project into staged artifacts.
Next Step
Read the product shape, then open the real app.
Start with the overview if you want the full repository context. Jump into the app if you already want to inspect the workspace, scan surface, and review flow.