📖About

The Activity Tracker's done-per-day chart and day drill-down hide every planner-mode initiative: an activity-count stamp is written only when a bypass closes, planner closes never receive one, and the tracker trusts the stamp alone - so autonomous pipeline deliveries read as zero forever. Fix reader-side per the derive-from-source doctrine: when a closed initiative has no stamped activity count, the tracker derives it live - first from the agent activities ledger, then from the planner execution book rows recorded for that initiative (normalizing the mixed id formats stored there). Applied to both the daily chart and the day drill-down so they agree. Second defect in the same endpoint: the RSI-retriggered exemption (closed RSI-loop work should count even with zero logged activities) is computed but never applied to the filter, and its lookup runs on an already-closed database handle so it always comes back empty - move it inside the connection scope and apply it in both endpoints. Junk closes with no recorded work anywhere stay hidden.

⚖️Decisions

• Elrond's bypass methodology checklist (a reminder, not a gate -- you've got this): 0. File it requested_mode='bypass' (planner-vs-bypass is a deliberate choice). bypass_start REFUSES a non-bypass initiative (CAROL-INI-1846), and the dispatcher only skips the bypass lane when the mode says bypass -- a 'planner' mistag lets Merlin's pipeline grab the placeholder step and block your finished work. 1. Filed as planned status -- let the bypass claim/activate it; never file active. 2. Open the bypass (bypass_start) with your droid id + the remediation answer (remediates_initiative_id=NNN, or remediates_nothing=True). 3. Work the blocks for your work-type: template -> design -> code -> test -> review. Do the real work; record decisions on the initiative as you make them. 4. Reality is recorded for you at close -- code (files changed), each decision, and the twin-review verdict become real activities tied to this initiative and show in the Activity Tracker like a planner run (CAROL-INI-1840). No dummy rows. 5. Keep the initiative status moving; it parks in 'reviewing' and is tagged uat-pending for you at close (CAROL-INI-1836), so the stuck-watchdog leaves it alone until UAT. 6. Close runs the gates (design/architecture compliance + caller-audit). If a gate flags something pre-existing or unrelated to your change, waive it with a clear written rationale -- audit, don't skip. 7. Bypass skips the planner's auto-orchestration, NOT the standards. Same template checklist, same review, same observability as a planner run. (elrond)
• [status-router] planned -> executing | event=bypass_executing | bypass transition (or-bx-01)
• Fix reader-side, not core-side: the planner execution book already records every planner-lane run, so the tracker derives activity live from it (derive-from-source doctrine) instead of adding a stamping hook at the close transition, which lives in the OS-locked pipeline core. This also heals all history automatically with no backfill. (orion)
• Derivation order: the agent activities ledger wins when it has rows; the execution book is the fallback (normalizing the legacy prefixed id format). Stamped bypass metrics keep priority in the daily chart, and the drill-down falls back to the stamp for parity, so bypass behavior is unchanged. (orion)
• The RSI-retriggered exemption is applied in both the daily chart and the day drill-down, and its tag lookup was moved inside the open connection scope - it previously ran on a closed handle, silently returning empty every time. (orion)
• Id lists are chunked at 500 in the derivation query to stay under the SQLite variable limit that previously caused a silent full-table fallback elsewhere. (orion)
• [delivery-check] 3 pending must-have criteria stamped met at bypass_end on live re-performance evidence (CAROL-INI-3020): test test_ini3320.py: PASS (1 passed in 0.73s) (orion)
• [status-router] executing -> reviewing | event=bypass_reviewing | bypass transition (or-bx-01)
• [agt_028.mind] dispatched my droid ca-s2 (sequence next): Run the next on-demand compliance audit tranche over the four new reviewing items, risk-class first. Reporting-integrity tranche: CAROL-INI-3320-00 (Activity Tracker deriving planner activity from the execution book with an RSI exemption — verify the exemption is a documented policy carve-out, not an unaudited blind spot: exempted spend must still be visible somewhere auditable) and CAROL-INI-3322-00 (one initiatives-of-the-day number with RSI diagnosis spend 'clubbed and never counted' — verify that never-counted does not mean never-recorded; ledger completeness must survive presentation simplification). Definition/measurement tranche at lower severity: CAROL-INI-3323-00 (the 4-digit family becomes the standard initiative definition in the SSTs — check the redefinition is reflected consistently across SSTs and does not retroactively distort audit trails keyed to old identifiers) and CAROL-INI-3321-00 (Token Cost Tracker per-service chart — verify 'complete per-service spend' claims ag — My standing lesson holds: when the reviewing queue turns over, prior audit coverage expires with it. Four initiatives (3320–3323) are new since my last tranche. Two of them touch reporting integrity — exemptions and 'never counted' spend are how expenditure quietly leaves the auditable record, and a tracker that self-defines what counts as activity is a control I must verify, not trust. ca-s2 is my on-demand auditor; this is squarely within my dispatch authority. (agt_028.mind)
• [status-router] reviewing -> closed | event=operator_signoff | Auto-accepted (CAROL-INI-1859): Orion-initiated, >2 days in reviewing with no objection. (el-srac-01)

✅Success criteria

• A closed planner-mode initiative that performed real build work appears in the Activity Tracker done chart and day drill-down with a non-zero activity count, with history covered automatically and no manual backfill. (must_have)
• Closed RSI-retriggered initiatives are visible in the done chart even when no activities were logged, so RSI-loop throughput is no longer hidden. (must_have)
• Auto-closed alarm shells and other closes with no recorded work anywhere remain hidden from the chart, so the chart does not inflate. (must_have)