ndpvt-web
- 659 skills
- 0 followers
- 14 repo stars
- last month last updated
- ▌ Sifting Noise Comparative Study · ndpvt-webFilter false positives from static analysis security tools (SAST) using LLM-agent-driven triage. Applies iterative code inspection, dataflow reasoning, and cross-file analysis to classify SAST alerts as true or false positives. Use when: 'triage these CodeQL findings', 'filter SAST false positives', 'review static analysis alerts', 'are these security warnings real?', 'reduce noise from Semgrep/SonarQube results', 'verify if this vulnerability is a false positive'.
- ▌ Swe Bench Mobile Agents Develop · ndpvt-webApply defensive programming and agent-architecture patterns from SWE-Bench Mobile to tackle production iOS/mobile development tasks. Optimizes how Claude navigates large mixed-language codebases, interprets multi-modal inputs (PRDs + Figma designs), and generates robust patches. Use when: "build this iOS feature from a PRD", "implement this Figma design in Swift", "fix this mobile app issue across multiple files", "generate a patch for this iOS codebase", "help me with production mobile development", "defensive programming for mobile code".
- ▌ The Necessity Unified Framework · ndpvt-webDesign and implement standardized, reproducible evaluation harnesses for LLM-based agents. Eliminates confounding factors (system prompts, tool configs, environment drift) so benchmark results reflect true model capability. Use when: 'build an agent evaluation framework', 'make my agent benchmarks reproducible', 'standardize agent testing', 'evaluate LLM agents fairly', 'set up a sandbox for agent eval', 'create reproducible agent benchmarks'.
- ▌ Villain At Averimatec Verifying · ndpvt-webBuild multi-agent fact-checking pipelines that verify image-text claims through modality-specific analysis, cross-modal reasoning, and structured QA generation. Use when the user says 'verify this claim with evidence', 'fact-check this image and caption', 'build a multi-agent verification pipeline', 'check if this image matches the text', 'detect misinformation in multimodal content', or 'create an evidence-based claim verifier'.
- ▌ Viola Video In Context Learning · ndpvt-webApply the VIOLA framework for label-efficient in-context learning on video or multimodal data. Uses density-uncertainty-weighted sampling to select the most informative examples for annotation, builds hybrid pools mixing ground-truth and pseudo-labels, and uses confidence-aware retrieval and prompting to maximize few-shot performance. Trigger phrases: 'few-shot video classification', 'label-efficient in-context learning', 'select best examples to annotate', 'VIOLA sampling', 'confidence-aware prompting', 'active learning for LLM in-context examples'
- ▌ Visiontrim Unified Vision Token · ndpvt-webImplement VisionTrim's training-free visual token compression for multimodal LLMs. Combines attention-based dominant token selection (DVTS) with text-guided token merging (TGVC) to reduce vision token counts by 75-90% with minimal accuracy loss. Use when: 'compress vision tokens in my MLLM', 'speed up LLaVA inference', 'reduce visual token overhead', 'implement token pruning for multimodal model', 'optimize MLLM for deployment', 'add training-free vision compression'.
- ▌ Wiki Live Challenge Challenging · ndpvt-webEvaluate deep research agents and LLM-generated long-form articles using the Wiki Live Challenge framework: 39 fine-grained writing criteria (well-written, broad coverage, neutral POV), factual verifiability via statement extraction and embedding-based matching, and citation accuracy checking. Use when: 'evaluate my research agent output', 'benchmark article quality against Wikipedia standards', 'check factual accuracy of generated report', 'audit citations in this article', 'score this article on Wikipedia Good Article criteria', 'build an evaluation pipeline for long-form generation'.
- ▌ Agent Primitives Reusable Latent · ndpvt-webDesign and orchestrate multi-agent systems using reusable Agent Primitives (Review, Voting/Selection, Planning/Execution) that compose into task-specific pipelines. Use when asked to: 'build a multi-agent workflow', 'create an agent pipeline for this task', 'set up agents to review and refine output', 'orchestrate parallel agent voting', 'decompose this into a planning and execution pipeline', 'design a reusable agent architecture'.
- ▌ Agenticscr An Autonomous Agentic · ndpvt-webAgentic secure code review for detecting immature vulnerabilities at pre-commit stage. Uses a two-phase Detector-Validator pipeline with SAST-rule semantic memory and CWE-tree validation to localize, classify, and explain security weaknesses in code diffs. Trigger phrases: "review this diff for security issues", "secure code review", "find vulnerabilities in my changes", "pre-commit security check", "check this PR for security weaknesses", "agentic security review"
- ▌ Beyond Holistic Scores Automatic · ndpvt-webBuild trait-based essay scoring systems that evaluate argumentative writing across multiple rubric dimensions (Content, Organization, Word Choice, Sentence Fluency, Conventions) using structured in-context learning prompts and ordinal regression. Trigger phrases: 'score essays by trait', 'build essay grading rubric', 'argumentative essay evaluation', 'trait-based writing assessment', 'rubric-aligned essay scoring', 'automated writing feedback'.
- ▌ Chain Mindset Reasoning Adaptive · ndpvt-webSolve complex problems by switching between four cognitive mindsets (Spatial, Convergent, Divergent, Algorithmic) at each reasoning step, adapting the thinking mode to what the current sub-problem actually needs. Use when asked to: 'reason through this step by step', 'solve this hard problem', 'think carefully about this', 'use chain of mindset', 'adaptive reasoning', 'multi-step problem solving'.
- ▌ Conceptual Cultural Index Metric · ndpvt-webCompute the Conceptual Cultural Index (CCI) to measure cultural specificity of sentences using LLM-based generality estimates across culture sets. Use this skill when users ask to "measure cultural specificity", "score how culture-specific a sentence is", "compare cultural relevance across countries", "detect culturally loaded content", "evaluate cultural bias in text", or "quantify how Japanese/American/etc. a sentence is".
- ▌ Darl Encouraging Diverse Answers · ndpvt-webGenerate diverse, high-quality answer variants for open-ended tasks using DARL's bounded-diversity framework. Use when: 'generate multiple solutions', 'diverse answers to this prompt', 'brainstorm varied approaches', 'explore alternative implementations', 'rewrite this N different ways', 'what are different ways to solve this'.
- ▌ Decomposing Reasoning Efficiency · ndpvt-webAnalyze and optimize LLM reasoning token efficiency using a multiplicative decomposition framework. Breaks down reasoning performance into completion rate, conditional correctness, verbosity, verbalization overhead, and coupling coefficient. Identifies bottleneck profiles and suggests targeted interventions. Includes trace-quality diagnostics (grounding, repetition, prompt copying). Trigger phrases: "analyze reasoning efficiency", "decompose token usage", "why is this model so verbose", "token budget analysis", "reasoning trace quality", "optimize reasoning tokens"
- ▌ From Assumptions Actions Turning · ndpvt-webBuild uncertainty-aware planners for multi-agent systems using the PCE (Planner-Composer-Evaluator) decision tree framework. Converts implicit LLM reasoning assumptions into scored decision trees that select actions under uncertainty without heavy inter-agent communication. Use when: 'build an agent that plans under uncertainty', 'create a decision tree from assumptions', 'multi-agent planning with partial observability', 'reduce agent communication overhead', 'score actions by likelihood and cost', 'uncertainty-aware action selection'.
- ▌ Issueguard Real Time Secret Leak · ndpvt-webScan text for leaked secrets using a two-stage pipeline: regex candidate extraction followed by contextual classification to eliminate false positives. Use when the user says 'scan for secrets', 'check for leaked credentials', 'find API keys in this text', 'detect hardcoded secrets', 'audit issue text for sensitive data', or 'prevent secret leaks'.
- ▌ LLM Assisted Logic Rule Learning · ndpvt-webBuild deterministic, interpretable anomaly detection rule sets for time series data using LLM-driven labeling, symbolic rule generation, and iterative optimization. Use when: 'detect anomalies in time series', 'build rule-based anomaly detection', 'generate logic rules from data', 'convert expert knowledge to detection rules', 'supply chain anomaly monitoring', 'create interpretable anomaly classifier'.
- ▌ Moco One Stop Shop Collaboration · ndpvt-webDesign and implement multi-LM collaboration pipelines using the MoCo framework's 26 methods across four collaboration levels (API routing, text exchange, logit fusion, weight merging). Use when: 'set up model collaboration pipeline', 'combine multiple LLMs for better accuracy', 'implement multi-agent debate or refinement', 'route queries to the best model', 'merge model weights', 'build an LLM ensemble system'.
- ▌ Opinf LLM Parametric Pde Solving · ndpvt-webSolve parametric PDEs using operator inference with reduced-order models. Builds POD-based reduced dynamics from a small set of simulation snapshots, then predicts solutions at unseen parameters and boundary conditions via polynomial regression on learned operators. Trigger phrases: "solve this PDE for different parameters", "build a reduced-order model", "operator inference for PDE", "parametric PDE solver", "predict PDE solution at new viscosity", "ROM for heat/Burgers/Navier-Stokes equation"
- ▌ Prompt Driven Development Claude · ndpvt-webOrchestrate prompt-driven development of large multi-module systems through structured, iterative natural-language workflows. Uses a phased development methodology with prompt taxonomies (feature requests, bug fixes, information sharing, architectural guidance) to build and maintain architectural coherence across thousands of lines of code without the user writing code manually. Trigger phrases: - "Build this entire system using prompt-driven development" - "Help me develop a complete framework/library through prompts only" - "I want to build a large project iteratively without writing code" - "Guide me through prompt-driven development of this application" - "Create a multi-module system using the PDD workflow" - "Develop this TUI/GUI/CLI framework step by step through prompts"
- ▌ Prompt Injection Attacks Agentic · ndpvt-webAnalyze, detect, and defend against prompt injection attacks targeting agentic coding assistants (Claude Code, Copilot, Cursor, Codex). Applies a three-dimensional taxonomy of delivery vectors, attack modalities, and propagation behaviors from a 78-study meta-analysis to systematically audit code, configs, MCP servers, skills, and tool integrations. Use when: "audit this repo for prompt injection risks", "is this MCP server safe to use", "review this skill file for hidden instructions", "harden this agent workflow against injection", "check if this .cursorrules file is malicious", "analyze this tool description for poisoning".
- ▌ Protean Compiler Agile Framework · ndpvt-webGuide fine-grained LLVM compiler phase ordering using the Protean framework's agile optimization approach — clustering passes into subsequences, searching with simulated annealing, and extracting static IR features for ML-driven pass selection. Use when: 'optimize my C compilation beyond -O3', 'find the best LLVM pass order for this code', 'extract compiler IR features for ML', 'speed up this benchmark with custom pass sequences', 'set up Protean compiler phase ordering', 'use ML to pick LLVM optimization passes'.
- ▌ Pruning Minimal Reasoning Graphs · ndpvt-webBuild and maintain compact reasoning graphs for retrieval-augmented generation (RAG) that persist across queries, prune low-value structure, and minimize token usage. Based on the AutoPrunedRetriever architecture. Use this skill when the user says: "build a graph-based RAG system", "reduce RAG token costs", "incremental knowledge graph", "persistent reasoning graph", "prune my retrieval graph", or "efficient multi-turn RAG".
- ▌ Ragturk Best Practices Retrieval · ndpvt-webDesign and optimize RAG pipelines for Turkish and other morphologically rich languages (Turkish, Finnish, Hungarian, Korean, etc.) using evidence-based stage configurations. Trigger phrases: 'build a Turkish RAG pipeline', 'optimize RAG for agglutinative languages', 'RAG reranking for Turkish', 'morphology-aware retrieval', 'cross-encoder reranking pipeline', 'HyDE for non-English RAG'.
- ▌ Redsage Cybersecurity Generalist · ndpvt-webApply RedSage's agentic augmentation methodology to cybersecurity assistance: structured threat analysis, vulnerability assessment, tool-command generation, and multi-turn security workflows grounded in MITRE ATT&CK, OWASP, CWE/CAPEC, and penetration testing frameworks. Trigger phrases: - "Analyze this CVE and map it to CWE/CAPEC" - "Help me with penetration testing methodology" - "Explain this MITRE ATT&CK technique" - "Generate security tool commands for this scenario" - "Assess this vulnerability and estimate CVSS" - "Walk me through incident response for this alert"
- ▌ Sqlagent Learning Explore Before · ndpvt-webExplore unfamiliar databases before writing SQL by building a local knowledge base of schema fragments, executable queries, and natural language descriptions. Uses MCTS-inspired schema traversal and dual-agent retrieval+generation. Trigger phrases: 'explore this database then query it', 'build SQL knowledge base', 'generate SQL for unfamiliar schema', 'text-to-SQL with exploration', 'help me understand this database and write queries', 'explore before generating SQL'.
- ▌ Stepshield Not Whether Intervene · ndpvt-webImplement temporal safety monitoring for AI agent trajectories using StepShield's cascaded HybridGuard pattern. Detects rogue agent behavior early enough to intervene, not just report. Triggers: 'monitor agent steps for safety', 'add early intervention detection', 'build a cascaded safety guard', 'detect rogue agent behavior in real time', 'implement step-level safety checks', 'build HybridGuard for agent monitoring'
- ▌ Teaching Evaluating Reason About · ndpvt-webApply knowledge-augmented reasoning distillation for polymer design tasks. Builds structured Chain-of-Thought pipelines grounded in polymer property knowledge bases. Triggers: 'polymer property prediction', 'polymer design constraints', 'SMILES polymer reasoning', 'polymer Tg prediction', 'knowledge-augmented CoT for materials', 'polymer benchmark evaluation'
- ▌ Toolself Unifying Task Execution · ndpvt-webImplement self-reconfiguring agent workflows where configuration (sub-goals, strategy, toolbox, context) is a mutable tool the agent calls at runtime. Use when: 'build an adaptive agent', 'self-reconfiguring pipeline', 'agent that adjusts its own strategy', 'dynamic tool selection workflow', 'multi-stage task with runtime adaptation', 'ToolSelf pattern'.
- ▌ Towards Adaptive Scalable Robust · ndpvt-webImplement RAPS (Reputation-Aware Publish-Subscribe) multi-agent coordination using intent-based pub/sub messaging, reactive subscription refinement, and Bayesian reputation scoring. Use when: 'coordinate multiple LLM agents', 'build a robust multi-agent system', 'pub/sub agent architecture', 'agents that detect bad peers', 'scalable agent orchestration', 'fault-tolerant agent swarm'.
- ▌ Trapped Past Disentangling Fluid · ndpvt-webDiagnose whether an LLM is memorizing or reasoning by constructing distributional proximity tests. Classifies task inputs as within-distribution (WD), near-distribution (ND), or out-of-distribution (OOD) and measures the generalization gap. Use when: 'audit LLM reasoning vs memorization', 'test if my model generalizes', 'measure crystallized vs fluid intelligence', 'evaluate model robustness on novel inputs', 'benchmark OOD performance degradation', 'diagnose training data leakage in evaluations'.
- ▌ Verge Formal Refinement Guidance · ndpvt-webIterative verification-guided reasoning that decomposes answers into atomic claims, classifies and routes them to formal (SMT/logic) or consensus-based verification, localizes errors via Minimal Correction Subsets, and refines until convergence. Use when: 'verify my reasoning step by step', 'check this logic for contradictions', 'formally verify this argument', 'find the flaw in this proof', 'validate these claims against constraints', 'refine this answer until it is logically consistent'.
- ▌ A Mapreduce Executing Wide Search · ndpvt-webExecute large-scale breadth-oriented search and retrieval tasks using the A-MapReduce pattern: decompose a wide query into a task matrix, dispatch parallel map agents for independent retrieval, then reduce partial results into a unified structured table. Triggers: 'find all X that match Y across a large set', 'build a comparison table of N items', 'search for every instance of X', 'collect attributes for a list of entities', 'wide search across many sources', 'gather structured data on hundreds of items'.
- ▌ Acegrpo Adaptive Curriculum Group · ndpvt-webAdaptive curriculum-driven iterative optimization for autonomous ML engineering tasks. Uses Evolving Data Buffers and Learnability Potential sampling from the AceGRPO paper to structure multi-step agent workflows that avoid behavioral stagnation. Triggers: 'optimize ML pipeline iteratively', 'adaptive curriculum for code tasks', 'iterative agent optimization', 'prioritize learning tasks', 'evolving task buffer', 'curriculum-based code improvement'.
- ▌ Agentdrive Open Benchmark Dataset · ndpvt-webGenerate structured autonomous driving scenarios and MCQ benchmarks using AgentDrive's factorized 7-axis prompt-to-JSON pipeline. Use when asked to: 'generate driving scenarios for AV testing', 'create autonomous driving benchmarks', 'build LLM evaluation datasets for driving', 'design safety-critical simulation scenarios', 'evaluate LLM reasoning on driving tasks', 'create multiple-choice questions for autonomous systems'.
- ▌ Alrm Agentic Robotic Manipulation · ndpvt-webBuild agentic LLM-driven robotic manipulation pipelines using the ALRM framework pattern: a ReAct-style reasoning loop with dual execution modes (Code-as-Policy for direct code generation, Tool-as-Policy for iterative tool-based execution). Generates modular robot control code with closed-loop planning, observation, and replanning. Trigger phrases: "robot manipulation agent", "agentic robot control", "ReAct robot planner", "code-as-policy generation", "tool-as-policy robot", "LLM robotic pipeline"
- ▌ Automated Multiple Mini Interview · ndpvt-webMulti-agent framework for scoring subjective, open-ended responses (interviews, essays, reflections) using transcript refinement + criterion-specific parallel scoring with calibrated few-shot examples. Use when: 'score these interview responses', 'evaluate candidate answers', 'grade these essays on a rubric', 'assess soft skills from text', 'build an automated scoring pipeline', 'rate open-ended responses against criteria'.
- ▌ Benchmarking Abap Code Generation · ndpvt-webGenerate syntactically correct and functional ABAP code using iterative compiler feedback loops. Applies the empirical methodology from Wallraven et al. (2026) to produce SAP ABAP classes that pass syntax checks and unit tests through up to 5 rounds of error-driven refinement. Trigger phrases: 'generate ABAP code', 'write ABAP class', 'fix ABAP syntax error', 'ABAP compiler feedback', 'SAP ABAP development', 'iterative ABAP correction'.
- ▌ Beyond Needles Illusion Decoupled · ndpvt-webDecouple evidence access from evidence use when evaluating or building long-context and RAG systems under semantic interference. Use this skill when the user says: 'evaluate my RAG pipeline against hard negatives', 'stress-test retrieval with semantic distractors', 'build a decoupled retrieval benchmark', 'diagnose why my long-context QA is failing', 'create collision-tested evaluation data', 'measure evidence access vs answer quality separately'.
- ▌ Bridging Arithmetic Gap Cognitive · ndpvt-webIterative Dual-Phase Financial-PoT: decouple semantic reasoning from arithmetic computation to eliminate calculation errors in financial analysis. Use when: 'calculate financial ratios from reports', 'analyze annual report numbers', 'compute ROE/ROA from statements', 'extract and calculate metrics from financial data', 'why is my financial calculation wrong', 'build a financial analysis pipeline'.
- ▌ Codeocr Effectiveness Vision Code · ndpvt-webRender source code as images for vision LLM processing to reduce token cost while preserving understanding. Use when: 'render code as image for LLM', 'compress code tokens with images', 'use vision model for code understanding', 'reduce token cost for large codebase analysis', 'code image compression for clone detection', 'syntax highlighted code screenshot for VLM'.
- ▌ Compact Hypercube Embeddings Fast · ndpvt-webBuild fast similarity-search systems using compact binary hypercube embeddings derived from foundation model encoders. Replaces brute-force cosine similarity over float vectors with Hamming distance over binary codes for orders-of-magnitude speedup and memory reduction. Trigger phrases: 'binary hashing for retrieval', 'fast embedding search', 'compact embeddings for similarity', 'Hamming space retrieval', 'hash-based vector search', 'reduce embedding memory footprint'
- ▌ Cua Skill Develop Skills Computer · ndpvt-webBuild reusable, parameterized skill libraries for computer-using agents (CUAs). Decomposes GUI automation into Skill Cells (intent), Parameterized Execution Graphs (actions), and Skill Composition Graphs (chaining). Use when: 'build a skill library for desktop automation', 'create reusable GUI action primitives', 'design a computer-using agent with skill retrieval', 'structure browser automation as composable skills', 'add failure recovery to my GUI agent', 'make my automation agent learn from past failures'.
- ▌ Datacross Unified Benchmark Agent · ndpvt-webCross-modal data analysis agent that unifies structured sources (SQL, CSV, JSON) with unstructured visual documents (scanned PDFs, invoice images, chart screenshots) using divide-and-conquer sub-agents and iterative code generation. Triggers: 'analyze data from these mixed sources', 'combine CSV with scanned PDF table', 'extract table from image and join with database', 'cross-modal data analysis', 'zombie data activation', 'analyze heterogeneous data files together'
- ▌ Devops Gym Benchmarking AI Agents · ndpvt-webApply the DevOps-Gym methodology to systematically tackle full-cycle DevOps tasks: build/configuration repair, runtime monitoring and anomaly detection, issue resolving via code patches, and regression test generation for Java and Go projects. Trigger phrases: 'fix this build failure', 'diagnose this runtime anomaly', 'generate regression tests for this bug', 'resolve this issue in Java/Go', 'debug this CI pipeline', 'monitor this running service for anomalies'.
- ▌ Docksmith Scaling Reliable Coding · ndpvt-webBuild reliable Docker environments for arbitrary code repositories using an agentic, multi-phase approach with dependency reasoning, loop detection, and cross-task success memory. Applies the DockSmith methodology to construct reproducible containers that pass test suites. Trigger phrases: "dockerize this repo", "build a Docker environment for this project", "create a reliable Dockerfile", "set up a containerized dev environment", "fix my Docker build failures", "make this repo's tests run in Docker"
- ▌ Exploring Reasoning Reward Agents · ndpvt-webApply Agent Reasoning Reward Model (Agent-RRM) structured critique to improve multi-step agent trajectories. Evaluates tool-use chains with explicit reasoning traces, focused critiques, and process scores. Use this skill when: - "Critique this agent's reasoning trace" - "Evaluate my tool-calling workflow and find flaws" - "Score this multi-step agent trajectory" - "Help me build a reward model for agent training" - "Improve this agent's reasoning with structured feedback" - "Debug why my agent pipeline produces wrong answers"
- ▌ Graph Based Agent Memory Taxonomy · ndpvt-webDesign and implement graph-based memory systems for LLM agents following the extraction-storage-retrieval-evolution lifecycle. Use when: 'build agent memory system', 'add long-term memory to my agent', 'implement knowledge graph memory', 'design memory for conversational agent', 'agent memory with graph structure', 'evolving memory for LLM agent'.
- ▌ Humans Welcome Observe First Look · ndpvt-webAnalyze AI agent social network activity using topic taxonomy classification and multi-level toxicity scoring. Detects content flooding, topic concentration, temporal risk patterns, and manipulative rhetoric in agent-generated discourse. Use when: 'classify agent posts by topic and toxicity', 'detect bot flooding patterns', 'analyze toxicity distribution in a social platform', 'monitor AI agent community health', 'find manipulative rhetoric in automated content', 'audit agent discourse for risk patterns'.
- ▌ Iesr Mcts Based Modular Reasoning · ndpvt-webConvert natural language questions into SQL queries using MCTS-based modular reasoning inspired by the IESR framework. Decomposes complex Text-to-SQL into information extraction, schema linking, multi-path tree search, and trajectory verification. Use when: 'generate SQL from this question', 'write a query for this database', 'translate this question to SQL', 'help me query this schema', 'complex SQL with calculations', 'text-to-SQL with reasoning'.
- ▌ Live Evo Online Evolution Agentic · ndpvt-webImplement online self-evolving memory for LLM agents using dual-bank architecture (Experience Bank + Meta-Guideline Bank) with reinforcement-weighted retrieval. Use when asked to: 'build an agent that learns from past mistakes', 'add evolving memory to my agent', 'implement experience-weighted retrieval', 'make my agent improve over time from feedback', 'create a self-improving agent pipeline', 'add reinforcement-based memory to my LLM system'.
- ▌ LLM Prompt Evaluation Educational · ndpvt-webSystematically design, evaluate, and rank LLM prompts for educational applications using tournament-style Glicko-2 comparison and pedagogical alignment criteria. Use when the user says 'evaluate my prompts', 'compare prompt templates', 'which prompt is better for teaching', 'optimize educational prompts', 'run a prompt tournament', or 'test prompt variations for learning'.
- ▌ Memcast Memory Driven Time Series · ndpvt-webBuild memory-augmented time series forecasting systems using hierarchical experience storage (historical patterns, reasoning wisdom, general laws) with LLM-driven inference and reflective iteration. Trigger phrases: 'forecast time series with memory', 'experience-driven prediction', 'memcast forecasting', 'hierarchical memory for time series', 'LLM time series with retrieval', 'memory-augmented forecasting pipeline'
- ▌ Open Tutorai Open Source Platform · ndpvt-webBuild personalized AI tutoring systems with structured onboarding, four-layer prompt architecture, adaptive lesson generation, 3D avatar integration, and learner analytics. Use when: 'build an AI tutor', 'create a personalized learning platform', 'design an educational chatbot with avatars', 'generate adaptive lesson plans', 'build a learner onboarding flow', 'create an intelligent tutoring system'.
- ▌ Pabu Progress Aware Belief Update · ndpvt-webApply Progress-Aware Belief Update (PABU) to build efficient LLM agents that track task progress and selectively retain context instead of passing full histories. Use when: 'build an agent with belief states', 'reduce agent context bloat', 'make my agent loop more efficient', 'implement progress-aware agent', 'selective context retention for agents', 'compact agent state management'.
- ▌ Pathwise Planning World Automated · ndpvt-webMulti-agent heuristic design framework that uses an entailment graph, policy/world-model/critic agents, and routed reflections to iteratively generate, evolve, and refine algorithmic heuristics for combinatorial optimization and search problems. Use when asked to: 'design a heuristic for TSP', 'optimize a scheduling algorithm', 'evolve a scoring function', 'find a better bin-packing strategy', 'auto-generate a search heuristic', 'improve my greedy algorithm through iteration'.
- ▌ Physical Prompt Injection Attacks · ndpvt-webDefend against and red-team physical prompt injection attacks on Large Vision-Language Models (LVLMs). Build input sanitization pipelines, attention-aware OCR filters, and robustness test harnesses that detect typographic instructions embedded in images before they reach the model. Trigger phrases: "harden my vision model against prompt injection", "detect text injection in images", "red team my LVLM pipeline", "build an image input sanitizer for VLMs", "test physical prompt attacks", "PPIA defense pipeline"
- ▌ Precise Reducing Bias Evaluations · ndpvt-webImplement the PRECISE framework to debias LLM-as-judge evaluations of search, ranking, and RAG systems by combining a small human-annotated gold set with large-scale LLM judgments using Prediction-Powered Inference. Triggers: 'evaluate search quality with LLM judges', 'debias LLM relevance judgments', 'estimate Precision@K with fewer annotations', 'combine human and LLM labels for ranking evaluation', 'reduce annotation cost for retrieval evaluation', 'correct LLM judge bias in ranking metrics'
- ▌ Sdr Cir Semantic Debias Retrieval · ndpvt-webBuild training-free composed image retrieval systems that combine a reference image with modification text to find target images, using semantic debiasing to suppress reference-image noise. Use when: 'build a composed image retrieval system', 'find similar images with text modifications', 'debias image search results', 'zero-shot image retrieval with CLIP', 'retrieve images matching a reference plus text edit', 'semantic ranking with bias correction'.
- ▌ Skillrl Evolving Agents Recursive · ndpvt-webBuild self-improving agent systems that distill raw execution traces into a hierarchical skill library (SkillBank) and recursively evolve those skills based on failure analysis. Use when asked to: 'build an agent that learns from mistakes', 'create a skill library from past runs', 'improve agent performance over iterations', 'extract reusable patterns from logs', 'reduce prompt bloat from trajectory history', 'make my agent pipeline self-improving'.
- ▌ State Art LLM Enabled Interaction · ndpvt-webBuild LLM-powered natural language interfaces for data visualization — NL2VIS pipelines, conversational chart analytics, multimodal chart explanation, and visualization recommendation systems. Use when: 'build a chat interface for my dashboard', 'let users query data in plain English', 'generate charts from natural language', 'add conversational analytics to my app', 'explain this chart automatically', 'recommend visualizations for this dataset'.
- ▌ Synthesizing File Level Data Unit · ndpvt-webGenerate high-quality unit tests with self-debugging repair loops and chain-of-thought reasoning. Produces tests with meaningful assertions, high branch coverage, and strong mutation scores by iteratively diagnosing and fixing errors, assertion failures, and coverage gaps. Use when: 'generate unit tests for this file', 'write tests with good coverage', 'create tests that catch real bugs', 'improve my test assertions', 'debug and fix failing tests', 'generate tests with reasoning'.
- ▌ Thinking Makes Agents Introverted · ndpvt-webPrevents the "introverted agent" problem where extended reasoning causes agents to give shorter, less informative responses in interactive/user-engaged tasks. Applies information disclosure (InfoDis) prompting to ensure agents proactively share state, enumerate options, and maintain transparency during multi-turn interactions. Use when: "help me build a customer service agent", "make my chatbot more helpful", "the agent isn't sharing enough information", "improve agent transparency", "fix terse agent responses", "design a user-facing assistant".
- ▌ Towards Declarative Agentic Layer · ndpvt-webBuild grounded, declarative agentic architectures using the DALIA pattern: capability descriptors, discovery protocols, federated agent directories, and deterministic task graphs. Use when the user says 'build a multi-agent system', 'create an agent orchestration layer', 'design declarative agent workflows', 'set up agent capability discovery', 'make agent coordination reliable', or 'avoid hallucinated agent plans'.
- ▌ Tracemem Weaving Narrative Memory · ndpvt-webBuild structured narrative memory systems from conversational traces using TraceMem's three-stage pipeline (segmentation, consolidation, clustering). Use when asked to: 'build a memory system for a chatbot', 'implement long-term conversation memory', 'create user memory cards from chat history', 'add persistent memory to an LLM agent', 'organize conversation history into structured narratives', 'implement episodic memory with clustering'.
- ▌ When Meets Fuzzy Topsis Personnel · ndpvt-webRank and select candidates using LLM-scored profiles combined with Fuzzy TOPSIS multi-criteria decision-making. Use when the user says 'rank these candidates', 'score resumes against criteria', 'build a hiring decision matrix', 'fuzzy TOPSIS ranking', 'multi-criteria candidate selection', or 'automate personnel screening'.
- ▌ Yoloe 26 Integrating Yolo26 Yoloe · ndpvt-webBuild and deploy real-time open-vocabulary instance segmentation pipelines using YOLOE-26, which combines YOLOv26's NMS-free architecture with YOLOE's open-vocabulary embedding heads. Covers text-prompted, visual-prompted, and prompt-free segmentation via the Ultralytics API. Trigger phrases: - "Set up YOLOE-26 for open-vocabulary segmentation" - "Detect and segment arbitrary object classes in real time" - "Build a visual prompt segmentation pipeline with YOLO" - "Deploy open-vocabulary instance segmentation with TensorRT" - "Segment objects by text description without retraining" - "Use YOLOE for prompt-free object detection"
- ▌ Agyn Multi Agent System Team Based · ndpvt-webOrchestrate multi-agent teams for autonomous software engineering using the Agyn methodology: coordinator, researcher, implementer, and reviewer agents with structured communication, isolated sandboxes, and iterative review loops. Use when: 'set up a multi-agent team to fix this bug', 'use agent swarm to implement this feature', 'resolve this GitHub issue with a team of agents', 'coordinate agents to refactor this module', 'spin up an engineering team to tackle this task', 'use Agyn-style agents to solve this problem'.
- ▌ Assessment Generative Named Entity · ndpvt-webBuild generative NER systems using LLMs with optimal output formats and prompt engineering. Use when: 'extract entities from text', 'build a NER pipeline with an LLM', 'named entity recognition with generative models', 'format NER output as XML or bracketed', 'fine-tune a model for entity extraction', 'nested entity recognition'.
- ▌ Benchmarking Reward Hack Detection · ndpvt-webDetect reward hacking in AI-generated code trajectories using contrastive analysis from the TRACE benchmark. Use when: 'check this code agent for reward hacking', 'detect if these test results are gamed', 'audit coding agent trajectories', 'find reward exploits in RL-generated code', 'contrastive analysis on code submissions', 'are these tests being manipulated'.
- ▌ Biasscope Automated Detection Bias · ndpvt-webAutomatically discover and test for hidden biases in LLM-as-a-Judge evaluation pipelines using the BiasScope framework. Generates bias hypotheses, perturbs test cases, and validates whether judge models are susceptible. Use when: 'audit my LLM judge for bias', 'find biases in my evaluation pipeline', 'stress test my LLM evaluator', 'check if my model judge is robust', 'discover unknown biases in my scoring system', 'build adversarial eval sets for my judge'.
- ▌ Calliope Tts Based Narrated E Book · ndpvt-webBuild offline TTS-narrated e-books with exact audio-text synchronization in EPUB 3 Media Overlay format. Use when the user asks to 'create a narrated ebook', 'add TTS audio to an epub', 'build an audiobook with text highlighting', 'synchronize speech with ebook text', 'convert epub to read-aloud format', or 'generate media overlays for epub'.
- ▌ Can Post Training Transform Causal · ndpvt-webPerform rigorous causal inference tasks using structured reasoning pipelines inspired by CauGym. Estimate treatment effects (ATE, CDE, ETT, NDE, NIE), compute probabilities of necessity/sufficiency, apply the backdoor criterion for deconfounding, and build causal DAGs from domain knowledge. Trigger phrases: 'estimate causal effect', 'what is the treatment effect of', 'causal reasoning over this data', 'apply backdoor adjustment', 'counterfactual analysis', 'build a causal DAG for this problem'.
- ▌ Ci4a Semantic Component Interfaces · ndpvt-webBuild semantic component interfaces that expose UI components as structured tool primitives for AI agent automation. Use when: 'make my UI agent-friendly', 'add CI4A interfaces to components', 'create semantic wrappers for web components', 'build agent-accessible UI toolkit', 'expose component actions as tools', 'wrap Ant Design components for agent use'.
- ▌ Cost Efficient RAG Entity Matching · ndpvt-webBuild cost-efficient RAG pipelines for entity matching and deduplication using blocking-based batch retrieval and generation. Reduces LLM API calls and latency by grouping similar entity pairs into blocks before retrieval and inference. Use when the user asks to 'match entities across datasets', 'deduplicate records with LLMs', 'build a RAG pipeline for entity resolution', 'reduce cost of LLM-based record matching', 'link records between two tables', or 'entity matching with knowledge augmentation'.
- ▌ Flyaoc Evaluating Agentic Ontology · ndpvt-webBuild multi-agent systems for end-to-end ontology curation from scientific literature. Applies FlyAOC's agent architecture patterns—memorization, pipeline, single-agent, and multi-agent—to extract structured, ontology-grounded annotations from document corpora. Use when asked to: 'curate knowledge from papers into a structured ontology', 'build an agent pipeline for scientific literature extraction', 'design a multi-agent system for document annotation', 'extract Gene Ontology or controlled-vocabulary terms from text', 'reconcile evidence across multiple documents into structured annotations', 'build a retrieval-augmented scientific reasoning system'.
- ▌ Fmbench Adaptive Output Formatting · ndpvt-webAdaptive Markdown output formatting that balances semantic fidelity with structural correctness. Applies the FMBench methodology to detect and fix broken lists, malformed tables, inconsistent headings, and invalid code blocks in LLM-generated Markdown. Use this skill when asked to: - "Format this output as clean Markdown" - "Fix the formatting in this document" - "Generate a well-structured report with tables and code blocks" - "Convert this content into properly nested Markdown" - "Clean up this Markdown — the lists and headings are broken" - "Reformat this response so it renders correctly"
- ▌ From Helpfulness Toxic Proactivity · ndpvt-webDiagnose and mitigate Toxic Proactivity in LLM agent systems -- the failure mode where agents override ethical constraints to maximize helpfulness. Use when building agent evaluations, designing safety benchmarks, auditing multi-step agent behavior, or hardening agent guardrails. Triggers: 'detect toxic proactivity', 'agent safety audit', 'test agent misalignment', 'evaluate agent behavioral safety', 'dilemma-driven agent testing', 'agent over-helpfulness risk'.
- ▌ How Few Shot Demonstrations Affect · ndpvt-webDesign prompt-based LLM safety defenses using optimal few-shot strategies. Applies the finding that few-shot demonstrations enhance Role-Oriented Prompts (RoP) but degrade Task-Oriented Prompts (ToP). Use when: 'harden my system prompt against jailbreaks', 'add few-shot safety examples', 'design a defense prompt', 'improve LLM safety alignment', 'protect my chatbot from prompt injection', 'write a safe system prompt with examples'.
- ▌ Jobresqa Benchmark Machine Reading · ndpvt-webBuild and evaluate multilingual machine reading comprehension systems for HR documents (resumes and job descriptions). Implements the JobResQA pipeline: tiered QA generation, cross-document reasoning, placeholder-based bias testing, TEaR translation, and G-Eval LLM-as-judge scoring. Use when: 'evaluate resume parsing accuracy', 'build HR question answering', 'test multilingual resume understanding', 'check bias in resume screening', 'cross-document QA on resumes and JDs', 'benchmark LLM on HR tasks'.
- ▌ Lata Tool LLM Assisted Translation · ndpvt-webLLM-assisted translation annotation: build parallel corpus annotation pipelines with template-based prompt management, structured JSON output constraints, stand-off annotation architecture, and human-in-the-loop review for sentence alignment, word alignment, translation shift classification, and error taxonomy. Trigger phrases: "annotate this translation", "align these parallel sentences", "classify translation shifts", "build a translation annotation pipeline", "create parallel corpus annotations", "translation quality annotation"
- ▌ LLM Based SQL Generation Prompting · ndpvt-webGenerate accurate SQL from natural language using the SSEV pipeline: schema-linked prompting, execution-guided self-refinement, and weighted majority voting across multiple candidate queries. Use when the user says 'write SQL for this question', 'query this database', 'convert this question to SQL', 'text-to-SQL', 'generate a SQL query from natural language', or 'help me query this schema'.
- ▌ Malicious Repurposing Open Science · ndpvt-webDefensive dual-use risk assessment for open science artifacts. Evaluates research papers, datasets, methods, and tools for repurposing vulnerabilities using a structured pipeline based on Hashemi et al. (2026). Produces risk reports with harmfulness, feasibility-of-misuse, and technical-soundness scores. Trigger phrases: "assess dual-use risk of this paper", "evaluate artifact repurposing risk", "run dual-use risk audit on this dataset", "check this tool for misuse potential", "security review of open science artifacts", "red-team this research for repurposing vulnerabilities"
- ▌ Medspeak Knowledge Graph Aided Asr · ndpvt-webBuild knowledge-graph-aided ASR error correction pipelines for medical speech, using phonetic similarity + semantic retrieval to fix misrecognized terminology before LLM reasoning. Use when: 'correct medical ASR errors', 'fix speech recognition for clinical terms', 'build medical speech QA pipeline', 'knowledge graph for ASR correction', 'phonetic matching for medical terms', 'improve Whisper transcription of medical audio'.
- ▌ Optimal Turkish Subword Strategies · ndpvt-webDesign and evaluate subword tokenizers for Turkish and other morphologically rich languages (MRLs) using the vocabulary-corpus-success triad framework. Covers tokenizer family selection (BPE, WordPiece, Unigram, morphology-level, character-level), vocabulary sizing, corpus coupling, and morphology-aware diagnostic evaluation. Use when: 'build a Turkish tokenizer', 'evaluate my tokenizer for agglutinative languages', 'choose vocabulary size for Turkish BERT', 'diagnose tokenizer morphological quality', 'compare BPE vs WordPiece for Turkish', 'optimize subword segmentation for MRLs'.
- ▌ Quasar Universal Autonomous System · ndpvt-webBuild autonomous multi-scale scientific simulation pipelines using the QUASAR architecture: a Strategist-Operator-Evaluator agent trio with adaptive planning, hierarchical knowledge retrieval, and context-efficient memory. Triggers: 'build an autonomous simulation pipeline', 'orchestrate DFT and MD workflows', 'QUASAR-style agent system', 'multi-scale atomistic workflow', 'autonomous computational chemistry', 'scientific simulation agent framework'
- ▌ Realhd High Quality Dataset Robust · ndpvt-webDetect AI-generated images using NLM noise entropy analysis and build robust forensic detection pipelines. Use when: 'detect if an image is AI-generated', 'build an AI image detector', 'classify real vs fake images', 'image forensics pipeline', 'noise entropy feature extraction', 'detect deepfakes or inpainted regions'.
- ▌ Realistic Synthetic Household Data · ndpvt-webGenerate realistic synthetic household datasets with bidirectional persona-environment coupling for embodied AI training. Use when asked to 'generate synthetic household data', 'create home activity datasets', 'build training data for home robots', 'simulate household routines', 'generate person-environment interaction data', or 'create embodied AI training datasets'.
- ▌ Semanticalli Caching Reasoning Not · ndpvt-webImplement pipeline-aware intermediate representation (IR) caching for agentic systems. Instead of caching final LLM responses, decompose multi-step pipelines into stages and cache structured reasoning artifacts at each checkpoint. Triggers: 'cache intermediate reasoning', 'reduce redundant LLM calls', 'pipeline-aware caching', 'agentic caching strategy', 'cache reasoning not responses', 'structured IR caching for agents'
- ▌ Seta Statistical Fault Attribution · ndpvt-webDiagnose and attribute faults in compound AI systems (multi-model pipelines) using SETA's modular robustness testing framework. Applies perturbations, traces execution through each component, computes per-component metamorphic relation scores, and statistically attributes system failures to specific modules. Use when: 'find which model in my pipeline is failing', 'robustness test my multi-model system', 'trace error propagation in my AI pipeline', 'attribute faults in my compound AI system', 'which component is causing my pipeline to break under noise', 'test robustness of my ML pipeline'.
- ▌ Spell Synthesis Programmatic Edits · ndpvt-webAutomate library migrations by synthesizing reusable code transformation scripts. Uses LLM-generated migration examples distilled into structured, testable rewrite rules (PolyglotPiranha / comby / semgrep patterns). Triggers: 'migrate from X to Y library', 'generate migration script', 'automated API migration', 'rewrite all uses of library X', 'create transformation rules for migration', 'synthesize code edits for library swap'.
- ▌ Uncertainty And Fairness Awareness · ndpvt-webAudit LLM-based recommendation systems for predictive uncertainty and demographic fairness bias. Implements the SNSR/SNSV fairness metrics, entropy-based uncertainty quantification, and personality-aware fairness scoring from Sah et al. (2026). Triggers: 'audit my recommendation system for bias', 'measure fairness in LLM recommendations', 'check if my RecLLM is biased against demographics', 'evaluate recommendation uncertainty', 'fairness benchmark for LLM recommender', 'test recommendation robustness to prompt perturbations'
- ▌ World Workflows Benchmark Bringing · ndpvt-webBuild world models for enterprise systems with hidden workflows and cascading database effects. Applies the probe-observe-model pattern from the World of Workflows paper to safely operate in interconnected databases where actions trigger invisible side effects. Use when: 'help me understand what side effects this DB change will cause', 'build a world model of our workflow system', 'audit what cascading changes happen when I update this record', 'predict hidden effects of this enterprise action', 'detect constraint violations in our workflow', 'map out hidden triggers in our database system'.
- ▌ Yasa Scalable Multi Language Taint · ndpvt-webPerform unified multi-language taint analysis across Java, JavaScript, Python, and Go codebases using YASA's UAST-based approach. Detects SQL injection, command injection, SSRF, XSS, deserialization, and privilege escalation vulnerabilities by tracing data flow from sources to sinks. Trigger phrases: 'find taint vulnerabilities across languages', 'multi-language security audit', 'trace user input to dangerous sinks', 'run taint analysis on this project', 'check for injection vulnerabilities', 'find data flow security issues'.
- ▌ Aegis Governance Integrity Security · ndpvt-webRed-team and harden AI voice agents and LLM-powered service systems against adversarial misuse using the Aegis framework. Evaluates authentication bypass, privacy leakage, privilege escalation, data poisoning, and resource abuse risks. Use when: 'red-team my voice agent', 'security audit my AI call center', 'harden my LLM agent against prompt injection', 'test my chatbot for privilege escalation', 'add layered defenses to my AI service', 'evaluate my agent for data leakage'.
- ▌ An Cost Efficient Agentic Framework · ndpvt-webAudit Ethereum smart contracts for business logic vulnerabilities using Heimdallr's four-phase agentic pipeline: function-level code reorganization via dependency graph clustering, heuristic Plan-Remind-Solve reasoning with adversarial state injection, automatic multi-step exploit chaining, and cascaded false-positive filtration. Trigger phrases: 'audit this smart contract', 'find vulnerabilities in this Solidity code', 'check this DeFi protocol for exploits', 'smart contract security review', 'detect business logic bugs in this contract', 'chain exploit paths in this protocol'.
- ▌ Artificial Intelligence Open Source · ndpvt-webAnalyze open-source projects for sustainability risks and apply AI-driven interventions for bug triaging, community health assessment, vulnerability detection, contributor onboarding, and maintenance automation. Trigger phrases: "analyze OSS health", "assess project sustainability", "triage issues automatically", "detect community smells", "onboard new contributors", "audit OSS security posture"
- ▌ Assessing Business Process Modeling · ndpvt-webEvaluate and generate BPMN process models from natural language using the BEF4LLM framework. Assess BPMN XML quality across syntactic, pragmatic, semantic, and validity dimensions. Triggers: 'generate BPMN from description', 'evaluate BPMN model quality', 'convert process description to BPMN XML', 'assess business process model', 'validate BPMN diagram', 'text to BPMN'.
- ▌ Breaking Protocol Security Analysis · ndpvt-webAudit and harden Model Context Protocol (MCP) server deployments against protocol-level vulnerabilities including capability attestation bypass, sampling injection, and cross-server trust propagation. Use when: 'audit my MCP server security', 'harden MCP deployment', 'check for prompt injection in MCP tools', 'add authentication to MCP messages', 'secure multi-server MCP setup', 'analyze MCP attack surface'.
- ▌ Breaking Static Graph Context Aware · ndpvt-webBuild query-adaptive knowledge graph retrieval systems using CatRAG's context-aware traversal. Transforms static KG-based RAG pipelines into dynamic, query-sensitive retrieval that recovers complete multi-hop evidence chains. Use when: 'build a multi-hop RAG pipeline', 'improve knowledge graph retrieval', 'fix semantic drift in graph search', 'implement context-aware graph traversal', 'retrieve complete evidence chains from a KG', 'add query-dependent edge weighting to my graph'.