chain-of-density

提取 YouTube/Bilibili 字幕并生成结构化深度摘要 | Extract subtitles & generate structured summaries (Claude Code skill)

Chain-of-density study of 'Recursive Language Models' (Zhang, Kraska & Khattab 2025, MIT CSAIL, arXiv:2512.24601) - five-tier note, locator-verified. The prompt-as-environment paradigm Trellis implements: REPL offloading, recursive sub-calls, 10M+ token inputs.

Chain-of-density study of 'Better Harnesses, Smaller Models: Building 90% Cheaper Agents via Automated Harness Adaptation' (Yang, Zhao, Wu & Kästner 2026, CMU, arXiv:2607.08938) - five-tier note, locator-verified. SLM agents, meta-agent harness optimization.

Our chain-of-density note on 'From Sparse to Dense' (Adams et al. 2023, arXiv:2309.04269) — plus the note-taking methodology every OpenCnid paper repo follows. Sources stay canonical.

Chain-of-density study of 'Who Grades the Grader? Co-Evolving Evaluation Metrics and Skills for Self-Improving LLM Agents' (Zhang et al. 2026, arXiv:2607.12790) - five-tier note, locator-verified. Double Ratchet, drawback detectors, anchor discipline, Goodhart repair.

A system-mode chain-of-density map of Automattic/harper (Elijah Potter, Apache-2.0) — reverse-engineered from 4,460 commits and 2,266 PRs — centred on Weir, its declarative rule language where a rule and its own tests are one artifact. With SPARK-steering, judge, and complexity-convocation findings.

Chain-of-density study of 'Emotion Concepts and their Function in a Large Language Model' (Sofroniew et al. 2026, Transformer Circuits/Anthropic, arXiv:2604.07729) - five-tier note, locator-verified. Emotion vectors, steering, functional emotions, desperation-blackmail dial.

Chain-of-density study of 'Polymorphic Combinatorial Frameworks (PCF): Adaptive AI Agents' (Pearl, Murphy & Intriligator 2025, arXiv:2508.01581) - five-tier note, locator-verified. SPARK parameter spaces, topos theory, 1.25M Monte Carlo simulations.

Chain-of-density study of 'Verbalizable Representations Form a Global Workspace in Language Models' (Gurnee, Sofroniew, Lindsey et al. 2026, Transformer Circuits/Anthropic) - five-tier note, locator-verified. The Jacobian lens, the J-space, alignment auditing, counterfactual reflection training.

A practical portfolio and meta-analysis of advanced prompt engineering. This project not only demonstrates core techniques (ToT, CoD, Prompt Chaining) but also documents the development of a 'Master Workflow' designed to overcome inherent AI behavioral biases and limitations. Includes interactive Streamlit app with prompt playground.

📚 Notes, concepts & exercises from a Prompt Engineering course — covering LLMs, zero-shot, few-shot, chain-of-thought, RAG, and AI agent prompting techniques.