Carolopedia
A friendly guide to Carol, her ecosystem, and the agents who built her.
📖About & Usage
About
The RSI Dashboard is Carol’s single place for viewing every RSI measure—indicators used to follow how individual services are performing and changing over time. Instead of making people hunt through four separate apps, it brings the whole picture together in one consistent, easy-to-scan workspace. A catalogue-style opening page lists the available services, while each service page presents its measures as tabs.
Every measure uses the same trend layout, so readers can move between services without relearning the screen each time. This makes comparisons quicker and unusual changes easier to spot. The dashboard belongs to Carol’s quality-management area and is owned by Prometheus, Prometheus, the Head of Quality Management. Created on 9 July 2026, it replaced the four standalone RSI apps under initiative CAROL-INI-2230. It is a private app: signed-in access is required.
Usage Patterns
The RSI Dashboard is used whenever someone needs a broad quality overview or wants to investigate the history of one particular measure. A reader might begin on the service catalogue, choose a service that appears to need attention, switch among its measure tabs, and inspect the trends—all without leaving the dashboard.
For example, Prometheus might notice that one service’s results have shifted, open that service, and compare several measures in the shared trend view. Because each page follows the same pattern, the investigation stays pleasantly boring in the best way: less time deciphering interfaces, more time understanding what changed. Wider quality conclusions may then be reflected in the Quality Scorecard or supported by testing evidence in Argus's Tests & Results.
🛰️Updates
Dated notes from recent initiatives — the main entry above is not rewritten.
RSI diagnosis filings are not initiatives per the Ninad ruling; only the 4-digit family number defines an initiative. This corrects RSI Dashboard's categorization.
The Ninad ruling of 2026-07-23 changes how RSI Dashboard handles cookbook non-compliance: instead of autonomous self-heal, it must now file a REDIRECT initiative to Orion for manual handling.
As of 2026-07-16, fixed the broken 'for-X' title template for RSI diagnosis filings to correctly interpolate target names instead of showing raw initiative IDs, resolving duplicate clusters missed by the filing dup-gate.
Changed failure handling: a failed Albus bypass now leaves the target BLOCKED and passes it to the self-heal loop, capped at 3 attempts, instead of immediate abandonment. (2026-07-23) RSI Dashboard
The tracker now correctly honors the RSI exemption for planner-mode initiatives, so RSI-tracked activities are no longer hidden.
2025-03-07: RSI diagnosis spend is now clubbed to its target and no longer counted separately in the daily initiative tally. This affects the tracking logic in RSI Dashboard.
The RSI diagnosis loop now catches the StatusRouteError when a target exceeds the diagnosis cap, preventing a crash every minute. The loop reliably abandons the exhausted target.
The RSI Dashboard's diagnosis prompt now includes Scriber's analysis for the initiative under diagnosis. RSI Dashboard
The Al-BX droid now picks up planner initiatives that the RSI diagnosis self-heal loop abandoned, completing them on Claude Fable. This provides an automatic bypass for initiatives that would otherwise stall.
The un-abandon logic in the dispatch sweep no longer clears the abandon flag for hard-capped targets, stopping the infinite diagnosis loop that wrote repeated RSI files for capped targets. This directly affects RSI Dashboard.
The Dispatch Queue card is now mode-aware: in RSI mode it shows the next blocked initiative; in sprint mode it shows sprint-lane items. This improves visibility of the active queue for RSI Dashboard.
The RSI Diagnosis Loop failed at 2026-07-22 00:03:12, potentially causing stale data on the RSI Dashboard. Hermione is investigating and will re-trigger the loop. Hermione
Ninad ruling 2026-07-23: Escalation Queue now only shows operator-attention items (rsi-diagnosis-abandoned or escalated); workable blocked items move to the Dispatch Queue as next-in-line and upcoming in pick order. Cookbook 916 updated accordingly. RSI Dashboard
The RSI Pattern Diagnostician process (el-rsi-pattern-01) was detected as never having run and has been re-triggered by Hermione. The RSI Dashboard dashboard should now reflect updated pattern data.
A new blog and Carolopedia guide explain the RSI sprint mode and other pipeline statuses. This provides official documentation for the RSI Dashboard dashboard's operational model.
The dispatch breaker flag now stays in sync with the blocked initiative count, so the pipeline correctly forces RSI mode when an initiative is blocked. This ensures RSI Dashboard displays the accurate status.
The self-heal loop for RSI diagnosis is now learning-driven, stopping on plateau with a hard backstop instead of a fixed cap of three. This prevents blind repetition and wasted attempts. RSI Dashboard
RSI diagnosis attempts are now capped at three per blocked initiative, preventing runaway self-heal loops.
2025-04-01: Albus's RSI diagnosis no longer reads poisoned evidence from its own status-router spam, eliminating garbage verdicts. RSI Dashboard
The RSI Dashboard now correctly excludes operator bypass initiatives from the sprint alternation, preventing the autonomous lane from stalling. RSI Dashboard
Fixed a self-heal bug in RSI Dashboard where the retrigger path ignored concreteness tags, causing generic advice to loop instead of escalating to empty-advice handling.
The for-X title template in diagnosis filings is broken, causing duplicate filings for targets like RSI Dashboard to show raw initiative IDs instead of target names. A fix is planned to stop these per-detection duplicates and correct the title interpolation.
Success measure for RSI enforced to durable-pipeline-fix; prompts and code guards updated accordingly. See RSI Dashboard.
🗂️Tabs & Screens
Tab inventory is being built — see CAROL-INI-077 step 7.
👤Owner
Prometheus · Head of Quality Management📚Recent initiatives
Initiatives that touched this app — a short summary each; open one for the full story.