Local route check at generation time.
Observed work.
Named honestly.
This is the operating record for the ProDyn.ai agent. It shows recorded activity, live route state, work categories, and change over time. It does not pretend that volume equals value or that a formula can grade intelligence.
Distinct canonical run records in the 14-day window.
Canonical tool-result rows, not unique artifacts.
Runs containing at least one tool-result row.
Enabled scheduled jobs at generation time.
Work leaves a measurable wake.
Bars show daily tool-result rows. The green trace shows distinct run records. A spike means more recorded activity, not better work.
Two seven-day windows.
The comparison excludes the current partial day. Dates stay visible so an expert can see exactly what moved and what did not.
What the system touched.
These are event families, not capability grades. One operation may produce several events; the labels describe system contact, not business outcome.
Reads, searches, patches, and writes against working artifacts.
Commands, builds, tests, service checks, and process control.
Reusable skill reads, updates, and procedure management.
Tool events outside the named public families.
Rendered-page inspection, navigation, and browser QA.
Current-source searches and document extraction.
Image and screenshot interpretation.
Parallel delegated research and review activity.
Session recovery and durable-memory operations.
Explicit work-state and completion tracking.
Where the work entered and what shape it took.
Source counts are canonical. Work categories are broad rule classifications, useful for direction rather than forensic reconstruction.
Runs by source
Runs by type
The latest ten records.
Only UTC day, broad rule classification, and event counts appear here. No prompts, customer names, file paths, provider routes, or session keys.
What these numbers mean.
Operational telemetry is only useful when its limits are as visible as its totals.
Content-free operational event shape from this ProDyn.ai agent runtime.
Counts describe recorded work activity. They do not score intelligence, business value, success, or customer impact.
A tool-backed run contains at least one canonical tool-result row. It does not imply that the operation succeeded.
The public detail window is 14 calendar days. Daily compact capacity is 60 days, beginning when measurement started.
