thalarch-codebase-intellisted

# Thalarch Codebase Intel Do not “read the repo”. Build only the map needed for the task. ## 1. Scan order 1. repository rules and intent docs; 2. build manifests, language/toolchain evidence and module graph; 3. entry points relevant to the task; 4. inbound/outbound dependencies of the changed/broken feature; 5. data/control flow; 6. tests around the same behavior and its consumers; 7. CI/runtime integration points; 8. recent Git history for the affected surface. Every architectural/control-flow claim needs a concrete file path, command output, source location, or runtime observation. Mark important conclusions: - `FACT`; - `INFERENCE`; - `UNKNOWN`. Do not silently turn inference into fact. ## 2. Feature-level repair mapping When the request is “make this feature/module work” rather than one isolated symptom, map a bounded feature boundary before fixes: - primary entry points; - internal files actually participating in the path; - dependencies imported/called; - consumers/callers outside the feature; - config/env/schema/API contracts; - tests that directly exercise the feature and tests of important consumers; - recent changes in the affected surface. Do not mechanically read every file in a large folder. Follow the dependency/control-flow graph until the relevant boundary is understood. Use `thalarch-debug` for individual causal failures discovered inside that map. ## 3. Deliverable Create a task-focused context packet: - stack/languages/toolchain/modules; -