All categories
Code Review & Quality
Code review, refactoring, debugging, maintainability, and engineering quality
- 15,973
- 666
Security grades appear on each card once the skill has been scanned. Newly imported skills may briefly show without a grade until the backfill job runs.
Open in full browserBrowse code review & quality skills
Showing 10,729–10,752 of 15,973 skills
- Discreteness Diffusion LlmUnderstand fundamental limitations of applying diffusion to discrete text: position-agnostic corruption ignores linguistic structure, and token-wise training misses multi-token dependencies. Design text diffusion systems satisfying five essential properties: position-aware corruption, dependency-aware training, parallel consistency, linguistic structure respecting, and robust handling of token boundaries.Votes: 0GitHub stars: 6
- Diffusionvl Ar To DiffusionConvert pre-trained autoregressive vision-language models into diffusion VLMs without architectural modifications. Use block diffusion strategy enabling arbitrary-length generation and KV-cache reuse. Hybrid attention enforces bidirectional within blocks, causal between blocks. Requires less than 5% of data compared to prior diffusion VLM methods.Votes: 0GitHub stars: 6
- Diffucoder Diffusion CodeTrain masked diffusion models for code generation using coupled-GRPO to optimize non-autoregressive generation. Achieves 4.4% improvement on code benchmarks while reducing autoregressive bias.Votes: 0GitHub stars: 6
- Densegrpo Flow MatchingImprove diffusion model alignment by assigning step-wise rewards during denoising instead of terminal rewards. Fixes sparse reward signal mismatch in multi-step generation processes through ODE-based reward estimation.Votes: 0GitHub stars: 6
- Consistency Solver Preview RefineTrain adaptive ODE solvers that generate fast diffusion previews while maintaining consistency with full-step refinement. Learn context-aware integration coefficients through PPO without distilling base models. Achieve high-quality previews in few steps with 47% fewer steps than standard methods.Votes: 0GitHub stars: 6
- Coco Text ImageGenerates complex text-heavy and structured images by converting natural language descriptions into executable code that specifies layouts, then renders and refines. Achieves 68% improvement on structured image generation benchmarks.Votes: 0GitHub stars: 6
- Causal Concept Graphs Latent SpaceExtract sparse causal concept graphs from LLM activations using SAE and DAGMA, then validate through ablation to identify causally influential features. Bridges mechanistic interpretability with causal inference for understanding reasoning flow.Votes: 0GitHub stars: 6
- Causal Attention Lookahead KeysImplement CASTLE, a causal attention mechanism that dynamically updates key representations as context expands. Reduces validation loss by 0.006-0.037 across model scales while maintaining O(L²d) training complexity and O(td) decoding speed. Deploy for improved language model perplexity without inference overhead.Votes: 0GitHub stars: 6
- Automind Adaptive Data ScienceBuild LLM-driven data science agents grounded in empirical knowledge through expert knowledge base, tree search algorithms, and complexity-adaptive code generation, surpassing SOTA by 8% on MLE-Bench.Votes: 0GitHub stars: 6
- Attntrace Context AttributionIdentify influential texts in long contexts via attention weights using top-K filtering and context subsampling, achieving 10-20x speedup over perturbation methods.Votes: 0GitHub stars: 6
- Agent Data ProtocolStandardizes agent training data representation across diverse sources (API use, web browsing, coding, software engineering). Single lightweight protocol unifies 13 datasets enabling 20% performance gains without domain-specific tuning. Enables reproducible agent training and scalable data combination.Votes: 0GitHub stars: 6
- Agent Conductor Topology EvolutionOptimize multi-agent collaboration by learning task-specific interaction topologies. Use an LLM orchestrator to generate layered DAG topologies that adapt to inferred problem difficulty, treating agent interactions as a learned graph structure rather than fixed patterns.Votes: 0GitHub stars: 6
- Adversarial Flow ModelsCombines adversarial training with optimal transport constraints enabling deterministic, transport-optimal one-step image generation with FID 2.38 on ImageNet-256. Apply when you need fast generative models without teacher-student distillation complexity.Votes: 0GitHub stars: 6
- Paper2skill Systematic EmpiricismConvert systematic empiricism papers into ranked practitioner checklists. Extracts implementation tricks, hyperparameter findings, and design choice ablations with conditions of applicability. Use this skill when extracting skills from Category 4 (Systematic Empiricism) papers — '37 PPO details'-style papers, hyperparameter studies, or ablation-heavy guides that systematize scattered knowledge.Votes: 0GitHub stars: 6
- Paper2skill Scaling EfficiencyConvert scaling and efficiency papers into practical resource planning guides. Extracts empirical scaling laws, compute-optimal allocation rules, and budget-performance trade-offs. Use this skill when extracting skills from Category 11 (Scaling and Efficiency) papers — Chinchilla-style scaling law papers, Flash Attention efficiency papers, or knowledge distillation studies.Votes: 0GitHub stars: 6
- Workflow Review Changes[Workflow] Use when reviewing uncommitted changes that are cross-module, contract-changing or security-sensitive: specialist lenses, validated fixes, re-review until clean; --fix-loop adds an outer convergence loop. Focused diff: changes-review.Votes: 0GitHub stars: 3
- Why Review[Code Quality] Use when a workflow step or the user asks for rationale review: why needed/chosen, trade-offs, good idea or best practice. Only explicit rationale asks; not generic reviews or how-it-works. --fix-loop fixes/re-reviews.Votes: 0GitHub stars: 3
- Test[Testing] Use when a workflow step or the user asks for a local test run: runs tests and analyzes the summary report.Votes: 0GitHub stars: 3
- Sync Codex[Codex] Use when running the full Codex mirror sync and verify pipeline, or when `.claude/**` edits leave `.agents/`, `.codex/` or `AGENTS.md` stale.Votes: 0GitHub stars: 3
- Knowledge Review[Research] Use when a workflow step or the user asks for a knowledge artifact review: completeness, citations, confidence accuracy, template compliance.Votes: 0GitHub stars: 3
- Harness Setup[Quality] Use when a workflow step or the user asks for an agent quality harness: feedforward guides and feedback sensors.Votes: 0GitHub stars: 3
- Graph Export[Code Intelligence] Use when exporting the code knowledge graph. --format={json|mermaid}.Votes: 0GitHub stars: 3
- E2e Test[Testing] Use when a workflow step or the user asks for E2E test work: select, generate or update tests, or --mode=verify (run with runner evidence; --fix-loop, --visual-review={true|false}).Votes: 0GitHub stars: 3
- Custom Agent[AI & Tools] Use when creating, verifying, or enhancing Claude Code custom agent files.Votes: 0GitHub stars: 3