clarification

Template-guided Clarifying Question Generation for Web Search Clarification (CIKM'2021).

AskBench: LLM question-asking/clarification benchmark & dataset with evaluation and training code (paper: arXiv 2602.11199).

Public beta docs and examples for a pre-reasoning observability controller for LLM and agent workflows.

Structured and minimal ask_questions tool for Pi planning and clarification flows

`util.inspect` with additional type information

Udacity R based analysis of how the chemical contents of wind impacts quality.

Intent clarification skill and bilingual demo for turning vague human input into precise language people and AI agents can act on.

🧭 Stop guessing. Start building. A 95%-confidence requirement clarification gate for Codex and AI agents—ask less, learn more, execute safely.

DPO-trained decision layer for coding agents — raises pass@1 from 0.419 → 0.639 on frozen 14B generator (p<0.000001)

Private, local-first executive work orchestrator for macOS. Clarification-driven; runs local Ollama models overnight.

Interactive benchmark for evaluating LLMs on clarifying ambiguous code requirements. Paper: arXiv:2607.00711

MOBIUS INFINITY — governed answer-entitlement (MMV) + reflective questioning (RQA) with an OpenAI-compatible API. AGPL-3.0; patent pending.

uBITX - sketch (arduino C) for the raduino to work with the uBITX transceiver V4

Deterministic benchmark for whether tool-using agents clarify high-context Japanese requests.

Most RAG agents answer confidently about the wrong thing. This one asks first, only when evidence disagrees. Found and fixed a real retrieval bug: no stopword filtering meant 3/7 accuracy on natural questions, now 7/7.