Web UI Tour
A visual walkthrough of each screen. This is a tour, not the quickest onboarding path. If you just want to get running, start with Get Started in 2 Minutes.
Tour stops
1. Creating a team
Go to
/new
(or click "Create Team" in the nav).
You'll see a list of templates. Each one creates a team with agents and skills
already configured. Pick one and you're done.
Screen
Template picker
Fullstack
3 agentsFrontend, backend, and QA
Backend
3 agentsAPI development and data engineering
Security
2 agentsSecurity auditing and compliance
Click a template to create the team instantly.
2. Visibility
Teams are private by default. Only participants (machines that joined with an invite code) can view a private team's page. If someone without access visits the URL, they see a "This team is private" message.
You can toggle a team to public if you want anyone with the link to view and clone the config. Public teams show a read-only view of member names, roles, instructions, and skill IDs/descriptions. Knowledge and skill bodies are redacted. Invites, participants, and edit controls stay hidden from non-participants.
The visibility toggle lives on the team dashboard, right next to the team name.
Screen
Visibility toggle
When to use public vs. private
Use private for teams with proprietary instructions or internal workflows. Use public when you want to share your team setup with the community or let others clone your config without needing an invite.
3. Team dashboard
After creating a team, you land on the dashboard. This is where you manage members, skills, and invites. Invite links open a read-only preview — join via the CLI to get edit access.
Screen
Team dashboard
my-project
team_abc123
npx @teamrc/cli join trc_inv_a1b2c3
Members
3frontend-dev
Frontend development
backend-dev
Backend development
qa-engineer
Quality assurance
Skills
2code-style
all agentsCode Style Guide
write-tests
Testing Requirements
Click any member card to edit their instructions and skills. Click a skill to edit its body.
4. Adding a member
Click "Add team member" on the dashboard. You'll see the agent catalog, grouped by category. Pick a pre-built agent or create a custom one.
Screen
Agent catalog picker
Core Development
frontend-dev
Frontend development
backend-dev
Backend development
fullstack-dev
Full-stack development
mobile-dev
Mobile development
Quality
qa-engineer
Quality assurance
code-reviewer
Code review
Picking a catalog agent pre-fills their instructions and recommended skills.
After picking an agent, you see a confirmation with the name, role, and a preview of what's included. Click "Add" and the agent joins your team.
Screen
Add member confirmation
Name
Role
5. Editing a member
Click a member card on the dashboard to open their detail page. Here you can edit their name, role, and instructions. You can also toggle which skills are assigned to them.
Screen
Member detail
Name
Role
Instructions
Skills
2Click a skill to toggle it on or off for this member. Skills marked "all agents" are always active and cannot be toggled.
6. Adding a skill
Click "Add skill" on the dashboard. You can pick from the catalog or write one from scratch. Skills are defined at the team level, then assigned to individual members on their detail pages.
Screen
Skill editor
ID
Title
Body
The body is markdown. Write whatever instructions you want the agent to follow. Enable "Apply to all agents" for team-wide rules.
7. Sharing your team
There are two ways to share a team. They work differently.
Invite codes (sync access)
Invite codes create an ongoing sync connection when used via the CLI.
Running teamrc join <code>
makes a machine a participant that receives updates and can push changes back.
Opening an invite link in a browser shows a read-only preview of the team.
Invite codes expire after 24 hours.
Team owners can find the Invites section on the dashboard. Click "Generate invite" to create a new code. Share it as a CLI command or as a URL.
Screen
Invites section
Invites
Generate invitenpx @teamrc/cli join trc_inv_x7y8z9
copy
Expires in 24 hours
To open this team in a browser, visit:
/invite/trc_inv_x7y8z9
Clone tokens (read-only copy)
Clone tokens let someone
copy your team config. The clone supports read-only pull to fetch updates,
but does not grant push or sync access.
Clone tokens are only available for public teams. Unlike invite codes,
they do not expire. To fork a cloned team into their own, the user runs
teamrc init
which creates a new independent team on the relay.
When your team is public, the clone token and command appear on the dashboard. Visitors to your public team page also see a "Clone this team" section with the CLI command.
Screen
Clone token (public teams only)
Clone
publicnpx @teamrc/cli clone trc_cl_r4s5t6
copy
This copies the current team config. No ongoing sync. Use an invite code for that.
Clone vs. join
Clone
(
trc_cl_...
): one-time
snapshot, no sync, no expiry. Good for sharing templates or letting others try your setup.
Join
(
trc_inv_...
): ongoing
sync, full participation, expires in 24 hours. Good for teammates.
8. Connected machines
Your connected machines are listed on your
personal dashboard
at
/dashboard
, not on the team page. Only you can see your machines.
Other participants cannot see your machine names, hostnames, or tokens.
Each card shows the machine name (your computer's hostname), whether it's scoped to a specific
project or global, and when it last synced. Machine names only appear for machines that have
run
teamrc login
. The CLI warns you before sharing your hostname.
Screen
Dashboard: active machines
Active Machines 3
trc_ak_x7...q2
trc_ak_m3...k8
trc_ak_p1...w5
Privacy note
Machine names come from your computer's hostname. They are only shared when you
run
teamrc login
. The CLI warns you before sending it.
If a machine has not logged in, it shows as "Unnamed." Machines are only visible to you
on your personal dashboard. They never appear on the team page. Other participants
cannot see them.
9. Team knowledge
The knowledge section shows shared notes that your agents have written during their work. When an agent discovers something important (an architecture decision, a debugging insight, a gotcha), it appends to the team knowledge file. That knowledge syncs to every machine on the next pull.
Knowledge is managed through the CLI and your agents. You cannot edit it in the web UI. The dashboard shows you what's there.
Screen
Knowledge (populated)
Knowledge
## Database - Using Postgres 18, connection pool size is 10 - Migrations run automatically on deploy ## Auth - Ed25519 keypairs, one per machine - Tokens are base64url-encoded public keys - Never log private keys ## Frontend - Tailwind CSS with DaisyUI, no custom CSS files - All components in lib/teamrc_web/components/
Screen
Knowledge (empty)
Knowledge
No knowledge yet. Knowledge is managed via the CLI.
How does knowledge get written?
The
team-knowledge
skill handles automatic loading.
As an
alwaysApply
rule, it injects the knowledge file
into every agent's context at the start of each session and instructs agents to
append useful findings as
## <topic>
entries before finishing.
teamrc also installs built-in slash commands for manual control:
/trc-save-knowledge
to save a finding,
/trc-save-core
to save permanent preamble knowledge,
and
/trc-knowledge
to view and search knowledge mid-session.
These work in Claude Code and Cursor.
On
teamrc push
or
teamrc sync
, knowledge uploads to the relay.
On pull, every machine gets the merged result with duplicates removed.
When knowledge exceeds 100KB, the oldest
##
sections are pruned automatically.
The preamble (content before the first heading) is permanent and never pruned.
The full flow
teamrc init
(CLI)
teamrc pull
on your machine to apply changes
teamrc sync
whenever you want the latest