Dashboard: Agents
Agents List
Route: /agents (AgentList.tsx)
Lists installed agents. Row click goes to Agent Detail.
Agent Detail
Route: /agents/:id (AgentDetail.tsx)
Above the fold: Competencies, Skills, and Tools sections (counts and lists, not tabs) reflecting what's actually equipped/visible on this agent — see object-model.md for what these mean structurally (equipped_competencies, the derived mcp_servers allowlist, and the allow_skill_tools/allow_integration_tools visibility lists).
Below that, three tabs:
| Tab | Shows |
|---|---|
| Memory | The agent's KV/semantic memory entries (live.agentMemory) |
| Schedules | Cron jobs bound to this agent |
| Invocation History | Past invocations — source, duration, tokens, outcome |
A Live Console button links to /agents/:id/live (see below).
Live Console
Route: /agents/:id/live (LiveConsole.tsx)
A real-time view into an agent's active/recent execution sessions — session list, per-session execution events, and metadata, distinct from the historical "Invocation History" tab on Agent Detail (which shows completed invocations after the fact; Live Console is for watching one happen, or inspecting a recent session's event stream in detail).
Activation and lifecycle
Activate/suspend actions on this screen call the same gated activate/ungated suspend verbs documented in reference/cli/agent.md and architecture/object-model.md — there is no separate console-only lifecycle mechanism.
Related:
- architecture/object-model.md — what an agent owns, structurally
- reference/cli/agent.md — the CLI equivalent of activate/suspend/status
- reference/api/agents.md — the underlying API
- workflows.md — Run Detail links here from a step's
agent_id - catalog.md — where equipped competencies come from