Dashboard page map
The dashboard is the operator’s second surface. An owner mostly talks to the
chief in Telegram; the dashboard is where they go when a chat answer is not
enough — a list to scan, a run’s steps, a feature’s health, a request to
answer, a session to watch. Every page reads and writes through the same
engine ops an agent’s tools call, so whatever a page shows, an agent can
answer too — and can tell the owner where to look.
This page is for that second half: which page answers which of the owner’s
questions, and how to link one. The map below is generated from the
dashboard’s own route tree, so it matches the engine version this copy came
with. The dashboard’s capabilities as a list: features.md
§ Control Plane (Dashboard).
Link the owner with /go/<ref>
When you point the owner at something, link <dashboard URL>/go/<ref> — never
a page path from the map below. /go resolves the ref to the object’s page of
the day, so the link keeps working when a release moves pages; a page path is
for knowing where things are, not for links you write into a message.
- The dashboard URL is the deployment’s
DASHBOARD_PUBLIC_URL. An agent
reads it from endpoint_list: the inbound:engine:dashboard row’s url. A
null url means the deployment has no public dashboard — name the ref as
text (run:run_x) instead of a link.
- Brain code builds the same link with
dashboardRefUrl(ref) from
@omnislash-ai/config/dashboard-link (null without a public dashboard),
and a link to one view of an object with dashboardSubViewUrl(view).
- A ref is
<kind>:<id>, split on the first : (an id may hold more). In
the URL the : stays literal and each /-separated segment of the id is
percent-encoded (%, ?, #, spaces); a ref with a . or .. segment has
no link, because a browser resolves it away — name it as text.
- Shorter forms
/go also takes: a bare prefixed id (/go/run_…,
/go/req_…, /go/ses_…), and an exact id or name, which it looks up with
brain_search — one match opens it, several ask the owner to pick.
- One view of an object:
/go/cycle:<id>?proposal=<round> opens a round of
a self-evolving cycle, /go/role:<role>?episode=<runId> one chat episode of a
role. /go forwards no other query parameter.
A ref with nothing to open — a path no page owns, a planning session that
ended with no scenario, an id that no longer exists — gets an answer in place
that says so and offers to ask the chief.
| Kind | Id | Opens |
|---|
scenario | the scenario’s id, as its file declares it | its page under its feature |
heartbeat | the heartbeat’s id | its page under its feature |
hook | the hook’s id | its page under its feature |
role | the role’s name (system:chief for an engine role) | the agent’s page under its feature |
feature | the feature’s folder name | the feature’s page |
bot | the bot’s name | Bots |
chat | the chat’s id | Bots |
person | the person’s id | Bots |
run | run_…, pinv_…, exe_… or eval_… | the run’s page |
request | req_… | the request beside the Inbox |
file | a brain-root-relative path (constitution/models.json, memory/<line>/<path>) | the page that owns it: a constitution file’s page, a feature file’s entity (a scenario’s plan or pipeline module on its Definition), Knowledge for memory, Skills for a skill |
capability | the capability (chat:send) | Permissions, with it selected |
model | the model id | Agents › Models |
prompt | sp_… (a scheduled prompt) | its feature’s page with it marked, else its agent’s session |
watch | watch_… (a resource watch) | its feature’s page with it marked, else its agent’s session |
planning | the planning session’s id | its open clarification in the Inbox, else the scenario it plans for, else Runs while it is live |
session | ses_…, or an agent’s id | the session’s page |
cycle | cyc_… | the cycle on Evolution |
grant | grant_… | Permissions › Grants, with it selected |
tool | the tool’s name | Permissions › Tools, with it selected |
skill | the skill’s name | Knowledge › Skills, with it selected |
memory | <line> or <line>/<path inside it> | Knowledge, at that line or file |
connection | conn_… | Connections |
artifact | pub_… (a publication) | Knowledge › Artifacts, with it selected |
endpoint | <direction>:<type>:<name> (endpoint_list’s id) | Connections › Inbound, or the MCP tab |
check | the doctor check’s name | System › Doctor |
credential | the credential’s id (credential_list) | where its remedy is: Repos, Connections or System › Credentials |
provider | the LLM provider’s slug (openai) | Agents › Providers |
repo | the repo’s slug | Connections › Repos |
service | <type>:<name> | Connections |
widget | widget_… (a live widget) | no page of its own — the chip opens the dock, where widget_get answers it; the dock lists every open widget’s page and opens it in a dialog (widgets.md § 13) |
query | <query op>?<its input as a query string> (a view: /go/query:loop.policy) | the list page that renders that op |
Pages
By navigation item, in sidebar order: the item’s landing, its tabs, then the pages inside it
(one object each, <param> its id or name). “Answers” are the owner’s questions the page is
built to answer, in the owner’s words.
Home
| Page | Path | Answers |
|---|
| Home | / | What needs me right now? · What is running, what happened, what comes next? · Stop a plan in flight · How is the business doing (KPIs)? · What did my operator tell me this morning? Did the brief go out? · When does the engine itself run (brief, memory consolidation, loop analysis)? |
Inbox
| Page | Path | Answers |
|---|
| Inbox | /inbox | What needs me right now? · Can my operator answer for me while I’m away? Who is answering now? |
Features
| Page | Path | Answers |
|---|
| Features | /features | Is feature X healthy, and if not, why? · What is in feature X, and how does each thing get triggered? · Which other features break if I switch X off? · Find a heartbeat / hook / scenario by name across features · Find all heartbeats (or hooks, scenarios) across features · Change how the operator behaves brain-wide (brief time, retention, health thresholds) |
| Feature | /features/<feature> | Is feature X healthy, and if not, why? · Which other features break if I switch X off? · What are this feature’s agents doing? · What has my operator scheduled for itself (scheduled messages, watches)? · What is in feature X, and how does each thing get triggered? · When does the engine itself run (brief, memory consolidation, loop analysis)? · What is running, what happened, what comes next? |
| Agent | /features/<feature>/agents/<agent> | — |
| Context | /features/<feature>/agents/<agent>/context | What does agent A see before its first message? Which agents are over budget? |
| Definition | /features/<feature>/agents/<agent>/definition | — |
| Session | /features/<feature>/agents/<agent>/sessions/<session> | What did agent session S say and do? · Can I take over or steer a working agent? · Talk to my operator from here, in the same thread as Telegram |
| Session context | /features/<feature>/agents/<agent>/sessions/<session>/context | — |
| Archived conversation | /features/<feature>/agents/<agent>/sessions/<session>/conversations/<conversation> | — |
| Heartbeat | /features/<feature>/heartbeats/<heartbeat> | When will X fire next? Did it fire on time? |
| Definition | /features/<feature>/heartbeats/<heartbeat>/definition | — |
| History | /features/<feature>/history | What did my operator change on its own (this week, in feature F)? |
| Hook | /features/<feature>/hooks/<hook> | — |
| Definition | /features/<feature>/hooks/<hook>/definition | — |
| Scenario | /features/<feature>/scenarios/<scenario> | When will X fire next? Did it fire on time? |
| Blueprint editor | /features/<feature>/scenarios/<scenario>/blueprint | Edit a blueprint’s phases and tasks by hand |
| Definition | /features/<feature>/scenarios/<scenario>/definition | — |
Evolution
| Page | Path | Answers |
|---|
| Fleet | /evolution | Is X improving? What did evolution change? · What has evolution learned about X? · What did my operator change on its own (this week, in feature F)? |
| Methods | /evolution/methods | — |
Runs
| Page | Path | Answers |
|---|
| Runs | /runs | What is running, what happened, what comes next? · Why did run R fail, what did it do, what did it cost? · Stop a plan in flight · Is my pipeline edit live? Which runs did it break? · Which runs came from chat with no scenario? · What has my operator scheduled for itself (scheduled messages, watches)? · What is my operator watching or waiting for? |
| Run | /runs/<runId> | Why did run R fail, what did it do, what did it cost? |
| Cycle | /runs/<runId>/cycle | — |
| Evaluation | /runs/<runId>/evaluation | — |
| Files | /runs/<runId>/files | — |
| JSON | /runs/<runId>/json | — |
| Log | /runs/<runId>/log | — |
| War room | /runs/<runId>/sessions | What are a run’s agents doing right now, side by side? · Follow whatever is live in run X · Can I take over or steer a working agent? |
| Usage | /runs/<runId>/usage | — |
Agents
| Page | Path | Answers |
|---|
| Sessions | /agents/sessions | Watch N agents side by side and type into any · Start an agent of role R with model M in repo Y · Can I take over or steer a working agent? · What are this feature’s agents doing? · What did my operator send from my account / as me? |
| Agents | /agents | Which system agents does the engine run, which did I customise, and what do they cost? |
| Models | /agents/models | Which models route where, and are the providers up? |
| Providers | /agents/models/providers | Which models route where, and are the providers up? · Which tool circuits are open? |
Knowledge
| Page | Path | Answers |
|---|
| Memory | /knowledge | What’s in memory, and which skills exist? |
| Skills | /knowledge/skills | What’s in memory, and which skills exist? |
| Artifacts | /knowledge/artifacts | What is exposed publicly? |
Audit
| Page | Path | Answers |
|---|
| Cost | /audit/cost | What did X cost over 30 days? |
| Actions | /audit/actions | — |
| Budgets | /audit/budgets | Am I near my spend limit? Raise it |
Bots
| Page | Path | Answers |
|---|
| Bots | /chats | Who can use my bots, where, and with which agent? · Why didn’t the bot answer X? · Which groups is a bot in but not serving? · Is the bot’s Telegram setup healthy (privacy mode, admin rights)? · What did the bot see in a chat? |
| Identities | /chats/identities | — |
| History | /chats/history | — |
Connections
| Page | Path | Answers |
|---|
| Services | /connections | Which external services are connected and healthy? |
| Repos | /connections/repos | Which repos can agents work in, and is the brain synced? |
| Inbound | /connections/inbound | Which events reach my operator, from where? Re-emit one |
Permissions
| Page | Path | Answers |
|---|
| Capabilities | /permissions | What can my operator do without asking me? · What grants does X hold? |
| Tools | /permissions/tools | — |
| Tool stats | /permissions/tools/stats | Which tools fail or are slow? |
| Grants | /permissions/grants | What can my operator do without asking me? · What grants does X hold? |
System
| Page | Path | Answers |
|---|
| Overview | /system | Which external services are connected and healthy? |
| Doctor | /system/doctor | Is the config valid? What value does the engine actually use? · Is the bot’s Telegram setup healthy (privacy mode, admin rights)? · Is my pipeline edit live? Which runs did it break? |
| Credentials | /system/credentials | Is any credential expiring? |
| Context | /system/context | What does agent A see before its first message? Which agents are over budget? |
| Logs | /system/logs | What did the engine log for run R? |
| Event stream | /system/events | The raw event stream (debugging), live and filterable |
| Event types | /system/events/types | Which event types exist, and which reach no hook? |
| Terminal | /system/terminal | An agent is stuck and I need a shell now |
| Config | /system/config | Is the config valid? What value does the engine actually use? · Change how the operator behaves brain-wide (brief time, retention, health thresholds) |
| Data | /system/data | What is in the brain’s own tables and KV store? |
| Schedule | /system/scheduler | When will X fire next? Did it fire on time? · Is the scheduler armed, in what timezone, did reconcile run? · What did the scheduler just do (reconciled, skipped, planned)? · What has my operator scheduled for itself (scheduled messages, watches)? · When does the engine itself run (brief, memory consolidation, loop analysis)? |
Outside the navigation
| Page | Path | Answers |
|---|
| Atlas | /atlas | Show me the whole operation on one map |
| Request | /inbox/<request> | — |
By question
Each of the owner’s questions, and the pages that answer it.
- What needs me right now? —
/, /inbox
- What is running, what happened, what comes next? —
/, /features/<feature>, /runs
- Is feature X healthy, and if not, why? —
/features, /features/<feature>
- What is in feature X, and how does each thing get triggered? —
/features, /features/<feature>
- When will X fire next? Did it fire on time? —
/features/<feature>/heartbeats/<heartbeat>, /features/<feature>/scenarios/<scenario>, /system/scheduler
- Why did run R fail, what did it do, what did it cost? —
/runs, /runs/<runId>
- Is the scheduler armed, in what timezone, did reconcile run? —
/system/scheduler
- Find a heartbeat / hook / scenario by name across features —
/features
- Which other features break if I switch X off? —
/features, /features/<feature>
- What grants does X hold? —
/permissions, /permissions/grants
- What did X cost over 30 days? —
/audit/cost
- Is X improving? What did evolution change? —
/evolution
- What can my operator do without asking me? —
/permissions, /permissions/grants
- Which external services are connected and healthy? —
/connections, /system
- What did agent session S say and do? —
/features/<feature>/agents/<agent>/sessions/<session>
- What did the engine log for run R? —
/system/logs
- What’s in memory, and which skills exist? —
/knowledge, /knowledge/skills
- Which models route where, and are the providers up? —
/agents/models, /agents/models/providers
- What is exposed publicly? —
/knowledge/artifacts
- The raw event stream (debugging), live and filterable —
/system/events
- What did my operator change on its own (this week, in feature F)? —
/evolution, /features/<feature>/history
- Is any credential expiring? —
/system/credentials
- What does agent A see before its first message? Which agents are over budget? —
/features/<feature>/agents/<agent>/context, /system/context
- Is the config valid? What value does the engine actually use? —
/system/config, /system/doctor
- An agent is stuck and I need a shell now —
/system/terminal
- Who can use my bots, where, and with which agent? —
/chats
- Why didn’t the bot answer X? —
/chats
- Which groups is a bot in but not serving? —
/chats
- Is the bot’s Telegram setup healthy (privacy mode, admin rights)? —
/chats, /system/doctor
- Is my pipeline edit live? Which runs did it break? —
/runs, /system/doctor
- Can my operator answer for me while I’m away? Who is answering now? —
/inbox
- What are a run’s agents doing right now, side by side? —
/runs/<runId>/sessions
- Can I take over or steer a working agent? —
/agents/sessions, /features/<feature>/agents/<agent>/sessions/<session>, /runs/<runId>/sessions
- Watch N agents side by side and type into any —
/agents/sessions
- Follow whatever is live in run X —
/runs/<runId>/sessions
- Start an agent of role R with model M in repo Y —
/agents/sessions
- Edit a blueprint’s phases and tasks by hand —
/features/<feature>/scenarios/<scenario>/blueprint
- Stop a plan in flight —
/, /runs
- What did the scheduler just do (reconciled, skipped, planned)? —
/system/scheduler
- Which tool circuits are open? —
/agents/models/providers
- Am I near my spend limit? Raise it —
/audit/budgets
- What has my operator scheduled for itself (scheduled messages, watches)? —
/features/<feature>, /runs, /system/scheduler
- Show me the whole operation on one map —
/atlas
- Which tools fail or are slow? —
/permissions/tools/stats
- How is the business doing (KPIs)? —
/
- What did my operator tell me this morning? Did the brief go out? —
/
- Talk to my operator from here, in the same thread as Telegram —
/features/<feature>/agents/<agent>/sessions/<session>
- Which events reach my operator, from where? Re-emit one —
/connections/inbound
- Find all heartbeats (or hooks, scenarios) across features —
/features
- Which event types exist, and which reach no hook? —
/system/events/types
- Which repos can agents work in, and is the brain synced? —
/connections/repos
- What did the bot see in a chat? —
/chats
- Which system agents does the engine run, which did I customise, and what do they cost? —
/agents
- Change how the operator behaves brain-wide (brief time, retention, health thresholds) —
/features, /system/config
- What did my operator send from my account / as me? —
/agents/sessions
- What are this feature’s agents doing? —
/agents/sessions, /features/<feature>
- When does the engine itself run (brief, memory consolidation, loop analysis)? —
/, /features/<feature>, /system/scheduler
- Which runs came from chat with no scenario? —
/runs
- What has evolution learned about X? —
/evolution
- What is my operator watching or waiting for? —
/runs
- What is in the brain’s own tables and KV store? —
/system/data