Tirion Agent
Real-time monitoring dashboard for the Tirion autonomous AI agent. Shows agent state, recent activity events, and pipeline execution status as Tirion works through the OODA loop.
Page Overview
Tirion Agent
Monitor Tirion's current state, review the timestamped activity feed, and track active and recent pipeline runs
How to Use
Features
Agent Status
Live state panel showing whether Tirion is running, idle, or offline with an emoji status badge
Activity Feed
Timestamped event stream with icons for each event type — scans, recommendations, completions, and failures
Pipeline Summary
Active ADW pipeline runs triggered by Tirion, plus a history of recently completed runs per pipeline
OODA Loop
Visual representation of Tirion's Observe → Orient → Decide → Act cycle and its current stage
Agent Status
The status panel at the top of the page shows Tirion's current operational state.
| State | Badge | Meaning |
|---|---|---|
| Running | 🟢 Running | Tirion is actively scanning or executing a pipeline |
| Idle | 🟡 Idle | Tirion completed its last cycle and is waiting for the next scan interval |
| Offline | 🔴 Offline | Tirion process is not responding — may require investigation |
Tirion runs continuously on an Azure VM, scanning all registered codebases approximately every 2 minutes.
Activity Feed
The feed displays a reverse-chronological stream of timestamped events.
Event Types
| Icon | Event Type | Description |
|---|---|---|
| 🔍 | scan_completed | Tirion finished scanning a repository for improvement opportunities |
| 💡 | recommendation_generated | A new improvement recommendation is ready for review |
| ✅ | pipeline_completed | An ADW implementation pipeline finished successfully |
| ❌ | pipeline_failed | An ADW pipeline run failed — check the Pipeline Summary for details |
| 🚀 | pipeline_started | Tirion accepted a recommendation and triggered an ADW pipeline run |
| ⏭️ | recommendation_deferred | A recommendation was deferred and will be re-evaluated on the next cycle |
Pipeline Summary
The pipeline summary section shows Tirion-triggered ADW runs grouped by pipeline name.
Active Runs
Any currently-executing ADW pipelines are listed here with:
- Pipeline name
- Current step (plan / implement / commit / PR)
- Elapsed time
Recent Run History
A compact history of the last 10 completed runs per pipeline, showing state and duration. Use this to identify pipelines that are consistently failing.
Tirion's OODA Loop
Tirion operates an autonomous improvement loop:
- Observe — Scan registered codebase repos for patterns, anomalies, and improvement opportunities
- Orient — Generate structured recommendations with estimated impact and implementation complexity
- Decide — Wait for human acceptance via the Code Intelligence page (recommendations are never auto-applied)
- Act — Create a spec + digest task, then trigger the ADW pipeline to implement
Tirion never acts autonomously — every recommendation waits in a "pending" state until a human accepts or defers it from the Code Intelligence page. This page shows the outcome after acceptance; acceptance itself happens on the Code Intelligence page (coming soon).
Infrastructure
| Component | Details |
|---|---|
| Host | Azure VM (always-on) |
| Scan interval | ~2 minutes |
| Recommendation storage | wg_digests.tirion_recommendations |
| Event log | wg_digests.tirion_events |
| Pipeline trigger | ADW (adws/agent.py) via Palantiri sidecar |