Recorded operational metrics compiled from fleet.db agent_runs + domain tables. Shows real activity (runs logged, findings produced, integrations touched) rather than valuation estimates. Snapshot: --. Agents with no telemetry yet have not yet logged a run via the Step 6.5 data layer write — typically pre-AGENT.md scaffolds or agents that bypass the Bash sqlite3 pattern.

Click column headers to sort. 7d Cadence = runs logged per day over the last 7 days (oldest left, latest observation right; hover a bar for the date). 7d / Exp = those 7-day runs vs. the expectation parsed from the agent's schedule string — low values mean either a missed schedule or the agent isn't writing its Step 6.5 data-layer row (which is itself a useful signal). Source-health details are omitted from this public edition. Severity columns count findings tagged at that level. Self-rated quality uses legacy mixed scales; it is not a certified common 10-point score. Rows tagged (unregistered) have fleet.db runs but no entry in the agent registry.

Agent	Category	Runs	7d Cadence	7d / Exp	Days Active	Findings	Crit	High	Med	Source labels	Avg Q	Last Run

Click column headers to sort. Current Value = (SaaS + Time + Revenue) x Maturity Multiplier. Max = full values at Production stage (1.2x).

Agent	Category	SaaS Replace	Time Saved	Revenue	Raw Total	Maturity	Multi	Current $	Production scenario	Confidence	Priority

Analysis of agent overlap and recommended consolidations. Philosophy: don't remove agents — create aggregation layers above them where overlap exists. Keep specialized agents but reduce the number of independent dashboards to check.

Historical agent priorities are shown for the full included valuation inventory. Within each tier, open-item conviction is considered before absolute modeled value gap. Agents without an eligible open item remain visible.