Uncut

Under-the-radar AI repos, papers, and models that cleared Megadose Uncut — substance plus breakout, not the already-covered firehose.

Recent Uncut picks

  1. kuliantnt/qq-maid-bot
    #42 · repo · A local Rust service for a general-purpose QQ bot, tagged around OneBot11, low-memory use, RAG, and wxbot.
  2. noyce6983981-max/ocr-vlm-local-retrieval
    #40 · repo · A local-first prototype for OCR plus vision-language document retrieval, with an independent evaluation angle.
  3. avar6/GLM-5.3-Flash-BF16-gguf
    #41 · model · A GGUF BF16 quantization of GLM-5.3-Flash aimed at conversational and endpoint-compatible use.
  4. blackhaiyu-sudo/specrag
    #39 · repo · A Python evidence-based RAG knowledge base for PRDs, business rules, SOPs, process docs, and product screenshots.
  5. raiyanyahya/llmaker
    #38 · repo · A Go CLI for self-hosting a modern LLM stack from the terminal.
  6. goobolabs/somali-language-standard
    #37 · benchmark · A versioned, machine-readable Somali language standard spanning orthography, grammar, terminology, translation, and AI resources.
  7. K-Dense-AI/scientific-agents
    #36 · repo · A set of AGENTS.md profiles that encode expert scientific and engineering reasoning styles for AI agents.
  8. Agent-Field/reels-af
    #35 · repo · A Python multi-agent system for automating short-form video creation at a claimed low per-reel cost.
  9. YintongHuo/awesome-agent-trajectory
    #33 · benchmark · A curated collection of agent trajectory analysis techniques and benchmarks for studying how LLM agents behave over time.
  10. Krypto-Whitehat/qwen3.8-9b-uncensored-cyber-exploit-XRPL-v3
    #34 · model · A Qwen-derived text-generation model packaged for cybersecurity and XRPL bug-triage workflows.
  11. yuwen-cool/ywcrew
    #31 · repo · A TypeScript orchestration tool that dispatches tasks to locally subscribed AI coding agents in parallel.
  12. Baekpica/Qwen3.8-Flash-Next-Mixed-Quant-SSD-PLE-GGUF
    #32 · model · A GGUF mixed-quant version of Qwen3.8-Flash-Next for image-text-to-text use, tagged for SSD offload.
  13. Metadata-Aware Adaptation of a Generative Foundation Model for Conditional CMR Synthesis
    #26 · paper · Adapts a pretrained latent diffusion model to synthesize cardiac MRI conditioned on clinical metadata and slice position.
  14. Parameter-Efficient Self-Supervised Adaptation for EEG-FM under Fixed Computational Budgets
    #27 · paper · Tests whether updating only 9% of parameters can adapt EEG foundation models under fixed clinical compute budgets.
  15. Show HN: Mole – Deep research agent for your terminal
    #28 · repo · A terminal-based deep research agent that drew meaningful Hacker News discussion as a Show HN launch.
  16. Hilbert-beinghappy/seektty
    #29 · repo · A JavaScript terminal UI for DeepSeek Harness, aimed at making DeepSeek-based coding-agent workflows pluggable from the CLI.
  17. AliAkrami1375/Li-Translate
    #30 · repo · A Vue-based platform for AI subtitle generation and natural-language translation for video and audio.
  18. dondai44423/donsetch
    #21 · repo · A Rust web fetch, search, and crawl tool for AI agents that avoids API keys and external accounts.
  19. Recursive Experiential-Working Memory Evolution for Long-Horizon Agent Harnesses
    #22 · paper · An agent memory architecture that separates working memory from experiential memory to improve long-horizon skill selection.
  20. Appllama/appllama-skills
    #23 · repo · A collection of agent skills aimed at turning successful mobile app interaction patterns into native-quality screens.
  21. Johnson-Durui/Companion-Space
    #24 · repo · A local-first AI companion and study app combining self-hosting, RAG, voice AI, FastAPI, Next.js, and VRM topics.
  22. Rethinking Pre-Training and Augmentation for Zero-Shot Cross-City Object Detection
    #25 · paper · A study of pre-training and augmentation choices for object detectors that must generalize to unseen cities without target-data profiling.
  23. PhysMLLMs: Spatial Priors for Unified Referring Segmentation and Grounded Reasoning of Images and Videos
    #16 · paper · Injects spatial priors during training to reduce jitter, drift, and identity switches in video multimodal segmentation models.
  24. On-policy Distillation with Verifiable Reward
    #17 · paper · Combines on-policy distillation with verifiable rewards so LLM post-training gets both dense guidance and correctness feedback.
  25. Meta$^n$: Recursive Self-Improvement through Emergent Depth
    #18 · paper · Proposes recursive self-improvement for LLM agents by repeatedly applying a fixed meta-operation to the solver stack's own traces.
  26. ExpConCAD: Experience-Guided Text-to-CAD Generation from Shape Descriptions with Implicit Spatial Constraints
    #19 · paper · Infers missing spatial constraints in text-to-CAD generation by reusing construction experience from prior shape programs.
  27. Linear Probing Provides Robust and Efficient Detection of Machine-Generated Text
    #20 · paper · Shows simple linear probes can detect machine-generated text efficiently and hold up better out of domain than heavier detectors.
  28. Low-Rank Ternary Adaptation for Fine-Tuning Transformers
    #11 · paper · A fine-tuning method for ternary transformers that applies discrete ternary weight updates without dequantizing the base model.
  29. PeakBench: Benchmarking Resource-Aware Tool Invocation in LLM Agents
    #12 · benchmark · A benchmark for LLM agents that evaluates parallel tool calls under latency and resource-scheduling constraints.
  30. Syn2RealTrack: Bridging the Gap Between Synthetic and Real-World Datasets for Online Multi-View Multi-Target Tracking
    #13 · paper · A multi-camera tracking paper that splits the synthetic-to-real gap into calibration, shape-prior, and object-assumption issues.