Select a project or create a new one.
No knowledge files yet.
Click 🧠 Analyse project to have Claude explore the codebase and write CLAUDE.md + docs/knowledge/ — structure, nuances, and gotchas that every future session (and board card) loads automatically. Board cards also add to these files as they discover things while working.
Publish log
loading…
Git graph
loading…
Past conversations
Project config
Create a project in the Projects tab first — cards run against a project folder.
Work board
● auto-worker: per-agent limit
Inbox 0
Working 0
Needs attention 0
Pending review 0
Done 0
→
Create a plan to soundboard an idea.
Spec
The implementation card is built from this plus the transcript.Soundboard
New card
images go inline in the details · text/md files attach as chips
Card
Agents
PCs running the ClaudeLauncher agent app — projects are assigned to an agent, which runs Claude locally.
| Status | Name | Machine | Version | Claude CLI | Jobs | Max jobs | Usage (month / 5h) | Last seen |
|---|
No agents yet. Enroll one, then run the ClaudeLauncher agent app on that PC and paste in the hub URL + key.
Add API endpoint
Add container registry
Console
connecting…
File
loading…
Agent key
Paste this key into the agent app on the target PC. You can view it again later with Show key.
Select or create a preset.
Loading dashboard…
Dashboard
Jobs, AI time, tokens & cost — per day, per project, per agent.
→
Tokens per day
All projects — hover a bar for the in/out/cost breakdown.
Jobs completed per day
Board cards finished — hover for AI time spent.
Projects
Jobs, tasks and AI time from the board · tokens & estimated cost from transcripts.
| Project | Jobs done | Active | Tasks | AI time | Tokens | Tokens (all time) | Est. cost |
|---|
Agents
Card runs attributed to each agent PC.
| Agent | Jobs done | Running now | AI time | Tokens in | Tokens out | Est. cost |
|---|
Settings
Model lists per engine — these fill the model dropdowns on projects and cards. Neither CLI lists its own models, so you maintain them here.
Claude models
Model IDLabel
Codex models
Model IDLabel
Duplicate detection
Optional. Add an Anthropic API key and an AI compares each new card against similar existing ones and flags genuine duplicates before they run. No key = duplicate detection is off (nothing gets flagged).
No key set — duplicate detection is off.
Usage & results
Every duplicate check this key ran — its reasoning and result.
Email (SMTP)
Outgoing mail server used for platform emails — customer-portal lifecycle notifications and dev/portal invites. Admin only. No host set = outgoing email is off.
No host set — outgoing email is off.
Send a test email
Sends using the saved settings above — Save first if you just changed them.
API endpoints
Enterprise LLM APIs. Anthropic-compatible = full Claude CLI on agents (chat + cards, enterprise billing). OpenAI-compatible = chat only, called by the hub.
| Name | Type | Base URL | Default model | Key |
|---|
No endpoints. Projects run on Claude agents by default.
Container registries
Docker/OCI registries for container publishes. Connect one here (host + credentials, stored centrally), then pick it in a project's Container publish config. 🚀 Publish builds & pushes the image using these — no repo .pubxml needed.
| Name | Host | Username | Password |
|---|
No registries. Add one to publish container images from a project.
Documents
Reference docs & protocols. Assign a doc to projects so the AI can consult it, or attach one to a card as the source of truth for that task.
| Title | Projects | Size | Updated |
|---|
No documents yet. Add protocols, specs or reference notes here, then assign them to projects.
Customer support inbox
Bug reports and feature requests submitted from the customer portal.Select a ticket.
Customers
System users
Developers who use the hub. Until the first user is registered the hub is open; after that, sign-in is required. Each user is scoped to projects and granted action permissions.
| Name | Role | Projects | Permissions |
|---|