Skip to main content

Documentation Index

Fetch the complete documentation index at: https://iii.dev/docs/llms.txt

Use this file to discover all available pages before exploring further.

The console is the visual UI for iii systems. It lists workers, functions, and triggers, and shows traces, logs, and metrics from the iii-observability worker. For scripting or agent integration, call worker functions directly.

​
Launch the console

The console connects to a running iii engine, so start the engine first, then launch the console in a second terminal: 
# 1. Start the engine. Use --use-default-config for a scratch instance,
#    or run `iii` (no flags) to load ./config.yaml.
iii --use-default-config

# 2. In another terminal, launch the console.
iii console
Open http://127.0.0.1:3113 in your browser.
Naming review: iii console is verb-shaped (launches the console) but doesn’t follow the iii noun verb convention used elsewhere in the CLI. Flagged for naming review per project-rules/cli.md.

​
Workers page

Lists every worker process currently connected to the engine. Each row shows the worker’s name (or short ID), the project / framework / language it reported, a runtime badge with SDK version, an optional isolation badge (libkrun, docker, etc. as reported by the worker), IP, PID, the number of functions it has registered, in-flight invocations, and how long it has been connected. Selecting a worker opens a detail panel with full metadata, the project / framework / language fields the worker reported, live metrics (memory breakdown, CPU, event-loop lag, uptime), and the list of functions the worker has registered.
The isolation badge reflects what the worker self-reports via the III_ISOLATION environment variable. The libkrun launcher injects III_ISOLATION=libkrun automatically; for container or Kubernetes deployments, set III_ISOLATION=docker (or similar) yourself. Workers that don’t set the variable show no badge.

​
Functions page

Lists every registered function, grouped by namespace prefix (the part before ::). A toggle in the header includes or excludes system functions. Select a function to open its detail panel. The panel includes:
  • A description (pulled from the function’s request schema, if present).
  • A request-body editor pre-filled from the request schema as a template.
  • An Invoke button that calls the function with the editor’s payload.
  • A result view that shows the response (or error) with the call duration.

​
Triggers page

Lists every registered trigger with All / HTTP / Cron / Event / Other filter tabs that show counts per type. Each row shows a summary (HTTP method and path, the readable cron expression, or the event topic), a type badge, the function_id it routes to, and the function’s description. Selecting a trigger opens a detail panel with type-specific testing tools:
  • HTTP. Method dropdown (GET/POST/PUT/PATCH/DELETE), path parameter inputs, a query-parameter builder, a JSON body editor for write methods, and a Send Request button.
  • Cron. Schedule (human-readable + raw expression), next-run preview, status, and a Run Now button that fires the bound function immediately.
  • Event. Event topic display, a JSON payload editor, and an Emit Event button.
State-bound triggers don’t have a dedicated tester on this page; fire them by editing the relevant entry on the States page.

​
States page

Browser for the engine’s key-value state store. The layout is groups list (left), items table (center, sortable on Key / Type), and a detail panel (right) that opens on row click. Supported operations:
  • Add. Open the Add Item modal to write a new entry; persisted via state::set.
  • Edit. Update the value inline from the detail panel; fires any registered state:updated triggers.
  • Delete. Remove an entry from the detail panel; fires any registered state:deleted triggers.
Use the search bar to filter items by key; pagination handles large groups. The page does not push live updates over WebSocket; use the refresh control after external writes.

​
Streams page

Live WebSocket monitor for the messages flowing through the engine’s stream connections. The header surfaces inbound / outbound counters, total bytes, and latency. A subscriptions bar lists the streams you’re currently watching (subscribe and unsubscribe through the modal). Message rows show timestamp, stream name, event type, a truncated data preview, and message size. A direction filter and pause/resume controls let you isolate flows; selected messages open a detail panel with the full payload. The filtered set can be exported to JSON.

​
Queues page

Lists durable queue topics with Topic, Broker, a DLQ badge (when dead-lettered messages exist), and Subscribers columns. Selecting a topic opens a resizable detail panel with two tabs:
  • Overview. Live topic stats and a JSON publisher for sending test messages.
  • Dead Letters. Failed messages with retry and delete actions.
Counts and stats refresh by polling. Keyboard shortcuts: j / k to move through the list, Enter to select, 1 / 2 to switch tabs.

​
Traces page

OpenTelemetry trace visualization across four view modes:
  • Waterfall. Timeline view with spans laid out by start time and duration. Default.
  • Flame Graph. Stack-based view; wider bars mean longer-running spans.
  • Trace Map. Topology graph showing service-to-service edges via parent-child spans.
  • Flow. Node-based execution flow of parent-child span relationships.
Filter controls cover trace ID, service name, span / operation name, status (ok / error / unset), duration range, time range, and arbitrary span attributes; results can be sorted by start time, duration, or service. The selected span’s detail panel includes its name, service, duration, status, IDs, tags, logs, errors, and baggage.
Trace collection requires the iii-observability worker with traces enabled.

​
Logs page

Structured OpenTelemetry log viewer. Each row shows timestamp, severity badge (DEBUG/INFO/WARN/ERROR), a truncated trace ID, the source function / service, the message, and a context-field count badge. Expanding a row reveals the full attributes, resource metadata, and trace context. Filters cover severity, time range, and full-text search across message / trace ID / source. The trace ID on each row is clickable: it pivots the view to show only logs from that trace, which also lets you jump to the corresponding entry on the Traces page.
Log collection requires the iii-observability worker with logs enabled.

​
Configuration

The console reads engine and observability settings from CLI flags and environment variables. To list every flag the binary accepts (with its default), run: 
iii console --help
The corresponding environment variables share the same set; each flag’s help text names the env var that overrides it. The Config page in the sidebar shows the resolved values at runtime (the engine endpoints, ports, OpenTelemetry settings, and version the console is currently using), so you can confirm what’s actually in effect without re-reading the flags.