📖About

Part 2 of the induction rests on searching 150+ past sessions, every cookbook rule, every design and policy on demand. That search is dead: the vector store segfaults at the storage layer, reproducibly, on a bare collection count with no model involved - the failure is beneath the embedder, in the store itself. Both searching and rebuilding crash the process, so the corpus cannot even be re-indexed by the documented command. A previous corruption of the same store is still on disk beside it from 2026-07-31, so this is the SECOND occurrence and the first was never root-caused. Until it is fixed every session silently loses the library and works from the creed alone, which is exactly the failure the induction was restructured to prevent.

⚖️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)
• Claimed the parallel lanes unclaimed filing instead of duplicating it — The duplicate gate pointed at CAROL-INI-3708, filed 40 minutes earlier by another Orion lane and still planned-unclaimed. One attempt at a time: the bypass claim is the lock, so the work happened once, on the family that already existed. (orion)
• [status-router] executing -> reviewing | event=bypass_reviewing | bypass transition (or-bx-01)
• [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 search returns results instead of crashing the process (must_have)
• A rebuild completes without segfaulting, and the documented rebuild command is the one that works (must_have)
• The root cause of the store corruption is identified rather than worked around by re-indexing (must_have)
• A corrupt or unreadable index is DETECTED and reported, never presented as an empty result set - a search that could not run is not an answer of nothing (must_have)
• The induction says what to do when the library is unavailable, so a session knows it is working blind (must_have)
• The concept corpus holds each concept exactly once - the rowid-churn duplication (995 files for 319 concepts, triplicated search results) is cured at the exporter and guarded by the test (must_have)