User Guide¶
Welcome. I'm Loom, and this is how you and I work together.
This guide is for people who use me day-to-day -- creating projects, filing work, watching agents do their thing, and stepping in when I need a human decision. If you're looking to install or configure me, that's the Installation Guide. If you're the one keeping me running in production, see the Administrator Guide.
Getting In¶
Point your browser at http://localhost:8080 (or wherever your admin put me). Log in with the credentials they gave you. If nobody told you anything, try admin / admin -- those are the defaults, and yes, your admin should have changed them.
Once you're in, I hand you a JWT token behind the scenes. You won't notice it. The UI takes care of it.
The Dashboard¶
Here's what you'll see across the top:
| Tab | What I'm Showing You |
|---|---|
| Home | Your selected project -- its agents, its beads in kanban columns, its status |
| Kanban | Every bead across every project, with filters for when you need to find something specific |
| Decisions | Things I need you to weigh in on |
| Agents | Who's working, who's idle, who's assigned where |
| Personas | The roles my agents can take on |
| Users | User accounts and API keys |
| Projects | Everything I'm managing, with git status and controls |
| Providers | The LLM backends powering my agents, with health status |
| Conversations | Chat sessions between you and my agents |
| CEO | Your command center -- the big picture and a direct line to me |
| Analytics | How much work is getting done, how much it's costing, how fast things move |
What's in This Guide¶
- Dashboard -- How I lay things out and how you navigate
- Projects -- Adding repositories and bootstrapping from PRDs
- Beads -- My work items: tasks, bugs, features, and how they flow
- Agents & Personas -- The specialists I orchestrate and what they do
- Workflows -- Multi-step processes and how I drive them
- CEO Dashboard -- Your command center and the REPL
- Decisions & Approvals -- When I need your judgment
- Analytics -- Velocity, cost, and performance tracking
- Conversations -- Pair-programming with my agents
Common Questions¶
Beads aren't being picked up. What's happening?
A few things to check: Are agents assigned to the project? Is the project's git access working (readiness_ok: true)? Are the beads blocked by unresolved dependencies? If all of those look fine, check that at least one provider is healthy -- my agents can't think without an LLM.
A decision has been sitting there for days. I give decisions a 48-hour window by default. After that, I'll either escalate it or auto-resolve it depending on what kind of decision it is. If you want faster turnaround, check the Decisions tab regularly -- I put things there because I genuinely need your input.
How do I see what an agent is doing right now?
The Agents tab shows status. For real-time detail, you can stream events: curl -N http://localhost:8080/api/v1/events/stream?type=agent.status_change. Or just watch the bead cards move across the kanban -- that's the whole point of the UI.