fix(ui): chips populate input with example queries instead of auto-running

fix(ui+search): approachable starter prompts, animation timing, search robustness

docs(ui): privacy-first quick actions and starter prompts

fix(agent): remove content moderation from auditor

fix(agent): deterministic audit, synthesizer grounding, structured output reliability

fix(agent): thinker prompt injects real query, auditor nonsense-check tightened

docs: lib/knowledge browser RAG (IndexedDB, Xenova embeddings, retrieve flow)

docs(readme): deepen FAQ on agent boundary and LLM worker architecture

docs: env.example, README and ARCHITECTURE for web search and HITL

fix(agent): HITL resume via Command, cap reasoning, auditor nonsense filter, real web search

fix: simplify tools to web_search only and break auditor rejection loop

fix: enhance auditor's JSON parsing and error handling

fix: normalize auditor policy feedback

fix: improve auditor JSON handling and worker config

fix: use null resume for HITL stream

docs: update HITL and SSE docs

feat: add Hugging Face Hub CLI skill documentation

chore: lower max_new_tokens for SmolLM

chore: shorten HF short_description

Deploy: DeepTrust workspace (v2 clean history)