fix(ci): build Next.js UI before PyPI publish + simplify extraction prompts

style: ruff format + remove unused imports

Update test_consolidate.py

feat: prompt fixes

feat(dream): lightweight consolidation command (decay + dedup + cleanup)

feat(context): default coding agent context for extraction + recall

refactor(docs): split README into focused funnel + dedicated docs

feat(gate): extraction gate — smart throttling for LLM extraction

feat(search): deterministic memory-type mapping + temporal boost tuning

fix: pass TENSORY_DEBUG stderr through hooks, filter only uvx noise

feat: auto-refresh tensory package in plugin hooks

Update test_claude_code_llm.py

Update prompts.py

fix(timeline): larger slider hit area + document-level drag tracking

docs: add rule to never commit .claude/ directory