Quickstart
This walks the desktop app from first launch to a feed scored against your own stack. For the CLI path, see the MCP server.
1. Point 4DA at your code
On first launch, add the project directories you actually work in. 4DA's context engine (ACE) scans them — manifests like Cargo.toml, package.json, and go.mod, plus recent Git history — to learn your stack, your dependencies, and what you've been touching lately.
Nothing about your code leaves the machine. The scan runs locally and populates a SQLite database on disk.
2. Choose how you want AI (optional)
4DA works out of the box with private, on-device semantic search — no key required. You can go further:
- Local AI (fully private): point 4DA at Ollama for free local inference.
- Bring your own key: add an Anthropic or OpenAI key for AI-written briefings and deeper reranking.
You own the compute. 4DA never pays for it, never stores your keys remotely, and never makes a call you didn't configure. See Privacy & BYOK.
3. Run an analysis
Trigger a run from the top bar. 4DA fetches from your configured sources, scores everything against your context, and keeps only what passes the confirmation gate. Background monitoring can keep it fresh on an interval you set.
4. Read the three tabs
| Tab | What it answers |
|---|---|
| Brief | What are today's top picks, scored against my stack? |
| Preemption | What do I have to act on — CVEs, breaking changes, dependency risk — and what exactly do I run? |
| Signal | Which items earned their place through 2+ independent axes? |
Preemption opens on the Worklist: version-confirmed findings with their fix paths, and the Upgrade Plan. Two supporting views sit beside it: Blind Spots (dependencies whose coverage is thin) and Learning Opportunities (dependency news you haven't read yet).
5. Teach it
Save what's useful, dismiss what isn't. Saves build a preference profile you can inspect, pin, or forget; dismissals tell the Brief to stop showing you similar items. Relevance scoring itself stays grounded in your actual stack.
Next: The scoring engine.