all publishers

ndpvt-web

@ndpvt-web source repo

659 published skills · page 3 of 7

  1. ▌
    Can Implement Agent Based Odd Based · ndpvt-web
    Translate ODD protocol specifications into validated, executable agent-based model (ABM) code in Python. Use when the user says 'implement this ABM', 'convert ODD to code', 'build an agent-based model from this specification', 'replicate this NetLogo model in Python', 'translate this model description into a simulation', or 'create a predator-prey / ecological / social ABM'.
    14 repo stars
  2. ▌
    Can Small Handle Context Summarized · ndpvt-web
    Build context-summarized multi-turn QA systems that let small language models (SLMs) handle customer-service dialogues with near-LLM quality. Implements conversation history summarization, stage-based evaluation, and prompt engineering for resource-constrained deployments. Use when: 'build a customer service chatbot with a small model', 'summarize conversation history for context window', 'evaluate SLM vs LLM on multi-turn QA', 'deploy a multi-turn QA system on limited hardware', 'optimize dialogue context for small language models', 'stage-based analysis of chatbot performance'.
    14 repo stars
  3. ▌
    Dancing Chains Strategic Persuasion · ndpvt-web
    Apply Theory of Mind-based strategic persuasion to code reviews, PR rebuttals, RFC objections, and technical disagreements. Uses the ToM-Strategy-Response (TSR) pipeline from RebuttalAgent to model the reviewer's mental state, formulate a targeted persuasion strategy, and generate a grounded response. Trigger phrases: - "Help me respond to this code review" - "Write a rebuttal for this PR feedback" - "How should I push back on this review comment" - "Draft a response to this RFC objection" - "Help me persuade the reviewer" - "Respond to this technical critique"
    14 repo stars
  4. ▌
    Dynamic Role Assignment Multi Agent · ndpvt-web
    Dynamically assign specialized roles to multiple AI agents via a meta-debate protocol (proposal + peer review) before running the actual task. Use when: 'set up a multi-agent debate', 'assign agent roles dynamically', 'which model should handle which role', 'run a meta-debate to pick agents', 'optimize role assignment for my agent swarm', 'capability-aware agent selection'.
    14 repo stars
  5. ▌
    Epistemic Context Learning Building · ndpvt-web
    Build trust-aware multi-agent systems using Epistemic Context Learning (ECL). Constructs peer reliability profiles from interaction history so agents weight information by source credibility instead of blindly conforming. Use when: 'build a multi-agent pipeline with trust', 'add peer reliability tracking to my agent system', 'prevent sycophancy in my LLM agents', 'implement history-aware trust for multi-agent', 'make agents evaluate peer credibility', 'add epistemic context to agent orchestration'.
    14 repo stars
  6. ▌
    Eventcast Hybrid Demand Forecasting · ndpvt-web
    Build hybrid demand forecasting systems that fuse LLM-extracted event knowledge with time-series models using a dual-tower architecture. Use when asked to: 'forecast demand for flash sales', 'build a demand prediction pipeline with event awareness', 'integrate promotional calendars into forecasting', 'predict sales spikes from campaigns or holidays', 'combine LLM reasoning with time-series forecasting', 'create an event-driven demand model'.
    14 repo stars
  7. ▌
    Haif Human AI Integration Framework · ndpvt-web
    Apply the HAIF protocol to organize hybrid human-AI team workflows with tiered autonomy, delegation governance, and validation checklists. Use when: 'set up HAIF for our team', 'classify this task for AI delegation', 'generate a delegation registry', 'create AI validation checklists', 'add HAIF tier metadata to our sprint board', 'assess autonomy tier for this task'.
    14 repo stars
  8. ▌
    Identifying Concurrency Bug Reports · ndpvt-web
    Classify bug reports as concurrency-related using a four-level linguistic pattern taxonomy (word, phrase, sentence, report-level). Use when asked to 'triage concurrency bugs', 'find race condition reports', 'classify bug reports for threading issues', 'detect deadlock-related issues', 'filter concurrency bugs from issue tracker', or 'label threading bug reports'.
    14 repo stars
  9. ▌
    Interpreting Agentic Systems Beyond · ndpvt-web
    Audit and instrument agentic AI systems for system-level interpretability and accountability. Embeds traceability, causal analysis, and oversight mechanisms across the agent lifecycle—from goal formation through environmental interaction to outcome evaluation. Use when: 'add observability to my agent pipeline', 'trace why my agent made this decision', 'audit my multi-agent system', 'add interpretability logging to my LLM agent', 'debug compounding errors in my agent chain', 'instrument my agentic workflow for accountability'.
    14 repo stars
  10. ▌
    Latent Chain Of Thought As Planning · ndpvt-web
    Decouple reasoning from verbalization using PLaT-inspired latent planning. Maintains a broad solution space through parallel latent trajectories before committing to a single answer. Use when: 'explore multiple solution paths', 'latent reasoning', 'plan before answering', 'search diverse solutions', 'PLaT reasoning', 'breadth-first problem solving'.
    14 repo stars
  11. ▌
    Leveraging Turkish Skill Extraction · ndpvt-web
    Extract and normalize skills from job postings using a two-stage LLM pipeline: dynamic few-shot skill identification followed by embedding retrieval + LLM reranking against a standardized taxonomy (ESCO). Optimized for morphologically complex and low-resource languages like Turkish. Use when: 'extract skills from job descriptions', 'parse job postings for competencies', 'match skills to ESCO taxonomy', 'build a skill extraction pipeline', 'normalize skills from Turkish job ads', 'link extracted skills to a standard taxonomy'.
    14 repo stars
  12. ▌
    Medverse Reliable Medical Reasoning · ndpvt-web
    Decompose complex medical reasoning into DAG-structured parallel execution paths using Petri net theory. Improves accuracy up to 8.9% and throughput 1.7x over sequential chain-of-thought. Triggers: 'differential diagnosis reasoning', 'parallel medical reasoning', 'DAG medical analysis', 'structured clinical reasoning', 'MedVerse reasoning', 'multi-hypothesis diagnosis'
    14 repo stars
  13. ▌
    Mhdash Online Platform Benchmarking · ndpvt-web
    Build risk-aware evaluation pipelines for mental health AI assistants using the MHDash framework. Implements multi-dimensional annotation, risk-specific metrics, and multi-turn dialogue benchmarking that exposes safety-critical failure modes hidden by aggregate scores. Triggers: 'benchmark mental health AI', 'evaluate AI safety for mental health', 'risk-aware model evaluation', 'build mental health chatbot test suite', 'assess high-risk recall for AI', 'evaluate multi-turn dialogue safety'
    14 repo stars
  14. ▌
    Multi Agent Causal Reasoning System · ndpvt-web
    Build multi-agent systems that discover causal rules from event sequences using specialized agents (causal discovery, contextual information, orchestrator). Use when asked to: 'find causal patterns in event logs', 'build a multi-agent diagnostic system', 'automate rule discovery from sequences', 'generate Boolean fault rules from data', 'create an agent pipeline for causal reasoning', 'detect error patterns in time-series events'.
    14 repo stars
  15. ▌
    Neural Theorem Proving Verification · ndpvt-web
    Generate formal proofs for program verification conditions (VCs) in Isabelle, Lean 4, and Rocq. Translates C/WhyML code obligations into proof assistant syntax and synthesizes tactic-based proofs. Use when: 'prove this verification condition', 'generate Isabelle proof for this invariant', 'verify this C function formally', 'translate this VC to Lean', 'help me prove this loop invariant', 'synthesize a Rocq proof for this postcondition'.
    14 repo stars
  16. ▌
    Precision Practice Knowledge Guided · ndpvt-web
    Generate industrial-grade code summaries using the ExpSum knowledge-guided approach: function metadata extraction, domain term retrieval, function categorization, and constraint-driven prompting. Produces documentation that meets real developer expectations rather than generic semantic descriptions. Trigger phrases: - "Summarize this function/method/class" - "Generate code documentation" - "Write doc comments for this code" - "Document this API" - "Generate Javadoc/JSDoc/docstring" - "Improve these code summaries"
    14 repo stars
  17. ▌
    Predicting Intermittent Job Failure · ndpvt-web
    Classify and diagnose intermittent CI/CD job failures from execution logs using the FlaXifyer few-shot approach and LogSift log reduction. Trigger phrases: 'diagnose flaky build', 'classify CI failure', 'why did this pipeline fail intermittently', 'triage job log', 'find root cause in CI log', 'reduce CI log noise'
    14 repo stars
  18. ▌
    Privacy Collapse Benign Fine Tuning · ndpvt-web
    Audit fine-tuning datasets and pipelines for privacy collapse — the silent failure where benign training data degrades a model's contextual privacy reasoning while safety benchmarks stay green. Use when: 'audit my fine-tuning data for privacy risks', 'check if this dataset causes privacy collapse', 'evaluate privacy after fine-tuning', 'scan training data for privacy-degrading patterns', 'test model for contextual privacy norms', 'build a privacy-safe fine-tuning pipeline'.
    14 repo stars
  19. ▌
    Reasoning While Asking Transforming · ndpvt-web
    Proactive Interactive Reasoning (PIR) — instead of guessing when requirements are ambiguous or premises are missing, Claude pauses reasoning to ask the user targeted clarifying questions, then resumes with the new information. Produces more accurate code, math solutions, and document edits while using fewer tokens. Trigger phrases: "help me solve this (but I haven't given all the details)", "write code for this feature", "edit this document to match my intent", "figure out what I mean", "build this but ask me questions first", "reason through this problem with me"
    14 repo stars
  20. ▌
    Rubberduckbench Benchmark AI Coding · ndpvt-web
    Evaluate and improve AI coding assistant responses using RubberDuckBench's rubric-based methodology. Detects hallucinations, scores partial credit, and enforces truthful code reasoning. Use when: 'evaluate this code explanation', 'check my AI answer for hallucinations', 'score this coding response', 'audit code assistant accuracy', 'benchmark code Q&A quality', 'review this technical answer for correctness'.
    14 repo stars
  21. ▌
    Spider Sense Intrinsic Risk Sensing · ndpvt-web
    Implement event-driven, hierarchical security screening for LLM agent systems using Intrinsic Risk Sensing. Adds latent vigilance with adaptive defense escalation to agent pipelines. Use when: 'add security to my agent', 'defend agent against prompt injection', 'screen agent actions for risk', 'build a safe tool-calling agent', 'add intrinsic risk sensing', 'protect my LLM agent from attacks'.
    14 repo stars
  22. ▌
    Text Summarization Global Structure · ndpvt-web
    Summarize long documents while preserving global semantic structure and logical coherence using topology-guided pruning (GloSA-sum). Use when the user says: 'summarize this long document', 'compress this text for an LLM context window', 'extract key points while keeping logical flow', 'reduce this report but keep the reasoning chain', 'shorten this paper without losing structure', 'create a coherent summary of these meeting notes'.
    14 repo stars
  23. ▌
    Tokenomics Quantifying Where Tokens · ndpvt-web
    Analyze and optimize token consumption in LLM-based multi-agent software engineering workflows. Maps agent execution traces to SDLC stages (Design, Coding, Code Completion, Code Review, Testing, Documentation) and quantifies where tokens are spent. Use when: 'analyze token usage in my agent pipeline', 'where are tokens being wasted in my workflow', 'optimize my multi-agent system costs', 'profile my agentic coding pipeline', 'reduce LLM costs in my CI/CD agents', 'tokenomics analysis of my agent framework'.
    14 repo stars
  24. ▌
    Towards Automated Kernel Generation · ndpvt-web
    Automate GPU kernel generation and optimization using LLM-driven agentic workflows with profiling feedback loops. Use when user asks to 'write a CUDA kernel', 'optimize a Triton kernel', 'generate a GPU kernel for this operation', 'speed up this PyTorch operator with a custom kernel', 'convert this PyTorch op to Triton', or 'profile and optimize my kernel'.
    14 repo stars
  25. ▌
    Vistira Closing Image Text Modality · ndpvt-web
    Solve math problems from images by decomposing them into interleaved natural-language rationales and executable Python code, closing the image-text modality gap. Use when: 'solve this math image', 'extract and compute from this screenshot', 'OCR this equation and solve it', 'reason about this visual math problem', 'convert this handwritten math to code', 'help me with this homework photo'.
    14 repo stars
  26. ▌
    Whispers Wealth Red Teaming Googles · ndpvt-web
    Red-team LLM-based agentic payment systems against prompt injection attacks targeting transaction integrity and credential isolation. Use when: 'red-team my payment agent', 'test agent payment security', 'prompt injection audit for shopping agent', 'secure my AP2 implementation', 'harden agentic checkout flow', 'find vulnerabilities in my LLM agent pipeline'.
    14 repo stars
  27. ▌
    Whitespaces Dont Lie Feature Driven · ndpvt-web
    Detect whether source code was written by a human or generated by an AI (ChatGPT, Copilot, etc.) using whitespace, indentation, and stylometric feature analysis. Trigger phrases: 'is this code AI generated', 'detect machine generated code', 'check if code is human written', 'analyze code origin', 'code authorship detection', 'AI code detector'
    14 repo stars
  28. ▌
    Wideseek R1 Exploring Width Scaling · ndpvt-web
    Decompose broad information-seeking tasks into parallel subtasks using a lead-agent-subagent pattern with isolated contexts and result aggregation. Use when: 'research multiple competitors and build a comparison table', 'gather information about all X and summarize', 'find and compare pricing across providers', 'collect attributes for a list of items', 'build a structured overview from many sources', 'search for multiple things in parallel'.
    14 repo stars
  29. ▌
    Aacr Bench Evaluating Automatic Code · ndpvt-web
    Perform repository-level automated code review on pull requests using hierarchical context retrieval and structured defect classification. Triggers: 'review this PR', 'find defects in this diff', 'automated code review', 'review these code changes', 'check this pull request for issues', 'analyze this patch for bugs'
    14 repo stars
  30. ▌
    Accelerating Social Science Research · ndpvt-web
    Implement the EXPERIGEN agentic framework for automated hypothesis generation and empirical validation on datasets. Uses a Bayesian-optimization-inspired Generator-Experimenter loop to discover statistically significant, novel hypotheses from data. Trigger phrases: 'generate hypotheses from this dataset', 'discover patterns in social data', 'run EXPERIGEN on this data', 'automated hypothesis testing', 'find significant predictors in this dataset', 'data-driven hypothesis discovery'.
    14 repo stars
  31. ▌
    Adaptbpe General Purpose Specialized · ndpvt-web
    Adapt general-purpose BPE tokenizers into domain- or language-specialized tokenizers using the AdaptBPE post-training strategy. Replaces low-utility tokens with high-frequency domain-specific tokens to improve tokenization efficiency without retraining from scratch. Trigger phrases: "adapt tokenizer to domain", "specialize BPE for medical text", "optimize tokenizer for French", "reduce token fertility for code", "adapt vocabulary for legal documents", "domain-specific tokenizer"
    14 repo stars
  32. ▌
    Aligncoder Aligning Retrieval Target · ndpvt-web
    Repository-level code completion using AlignCoder's query enhancement and aligned retrieval technique. Generates candidate completions to build an enhanced query that bridges the semantic gap between unfinished code and the target completion, then retrieves precisely relevant cross-file context. Use this skill when the user says: - "Complete this code using context from other files in the repo" - "Help me finish this function using the rest of the codebase" - "Use cross-file context to complete this code" - "I need repo-aware code completion for this partial implementation" - "Retrieve relevant code from the repository to help complete this" - "What would go here based on how similar code works in this project?"
    14 repo stars
  33. ▌
    Birdturk Adaptation Bird Text To SQL · ndpvt-web
    Adapt Text-to-SQL systems and benchmarks for non-English, morphologically rich languages using controlled translation pipelines and agentic multi-stage reasoning. Triggers: 'translate SQL benchmark to Turkish', 'cross-lingual text-to-SQL', 'adapt BIRD dataset to another language', 'build multilingual SQL generation', 'text-to-SQL for low-resource language', 'non-English database querying'
    14 repo stars
  34. ▌
    Chain Simulation Dual Mode Reasoning · ndpvt-web
    Dual-mode reasoning framework that dynamically routes problems to specialized strategies: computational flow for math, symbolic JSON state tracking for spatial/entity reasoning, and hybrid fact-extraction for multi-hop inference. Use when asked to 'solve this step by step', 'reason through this problem', 'track state changes', 'figure out the answer to this logic puzzle', 'solve this math word problem', or 'chain these facts together'.
    14 repo stars
  35. ▌
    Chatting Images Introspective Visual · ndpvt-web
    Apply introspective visual thinking by iteratively 'chatting with images' — using language-guided re-examination of visual content to reason over fine-grained details, spatial relationships, and multi-image comparisons. Use when: 'analyze this image in detail', 'compare these images', 'reason about spatial layout', 'what's different between these screenshots', 'explain the visual relationship', 'trace the visual logic step by step'.
    14 repo stars
  36. ▌
    Comprehensive Comparison RAG Methods · ndpvt-web
    Select and configure the right RAG strategy for conversational QA systems based on dataset characteristics. Use when: 'build a conversational RAG pipeline', 'choose a RAG method for multi-turn QA', 'my RAG pipeline performs worse than no retrieval', 'optimize retrieval for dialogue systems', 'compare RAG strategies for my dataset', 'reranking vs HyDE vs hybrid BM25'.
    14 repo stars
  37. ▌
    Constrained Process Maps Multi Agent · ndpvt-web
    Build multi-agent workflows structured as constrained DAG process maps with Monte Carlo uncertainty estimation. Each agent occupies a specialized review role, with predefined escalation paths and terminal states (automated label or human review). Use when: 'build a multi-agent compliance pipeline', 'add uncertainty-aware escalation to my agent workflow', 'create a review chain with human fallback', 'design a DAG-based agent workflow with confidence thresholds', 'implement Monte Carlo sampling for agent decisions', 'build a multi-stage content moderation system'.
    14 repo stars
  38. ▌
    Cross Lingual Stability Judges Under · ndpvt-web
    Detect and fix cross-lingual evaluation instabilities in LLM-as-a-judge pipelines. Use when: 'audit my multilingual eval pipeline', 'check if my LLM judge is stable across languages', 'set up cross-lingual evaluation', 'calibrate judge scoring for non-English languages', 'diagnose ranking inversions in multilingual benchmarks', 'build controlled generation tests for eval reliability'.
    14 repo stars
  39. ▌
    Curiosity Driven Knowledge Retrieval · ndpvt-web
    Implements a curiosity-driven knowledge retrieval framework for autonomous agents. Formalizes agent uncertainty as a curiosity score, triggers external knowledge retrieval when uncertainty exceeds a threshold, and organizes retrieved knowledge into structured AppCards for selective integration into reasoning. Trigger phrases: "build an agent with curiosity-driven retrieval", "add uncertainty-aware knowledge lookup", "implement AppCard knowledge system", "create a curiosity-scored agent pipeline", "build adaptive knowledge retrieval", "implement uncertainty-triggered documentation lookup"
    14 repo stars
  40. ▌
    Darwin Dynamic Agentically Rewriting · ndpvt-web
    Evolutionary multi-agent code optimization using genetic algorithms. Agents mutate each other's training/configuration code, benchmark results, and select survivors across generations. Use when: 'evolve my training config', 'optimize this code with genetic search', 'set up evolutionary hyperparameter tuning', 'multi-agent code mutation pipeline', 'self-improving training loop', 'darwin-style evolutionary optimization'.
    14 repo stars
  41. ▌
    Decoupling Skeleton Flesh Multimodal · ndpvt-web
    Disentangled structure-content reasoning for table images and structured data. Separates table skeleton (layout/structure) from flesh (cell content) to answer questions accurately. Use when: 'analyze this table image', 'answer questions about this spreadsheet screenshot', 'extract data from this table photo', 'reason over this financial table', 'compare values in this table image', 'what does this table show'.
    14 repo stars
  42. ▌
    Egss Entropy Guided Stepwise Scaling · ndpvt-web
    Apply Entropy-Guided Stepwise Scaling (EGSS) to complex software engineering tasks like bug fixing, code generation, and refactoring. Uses entropy-based uncertainty detection to selectively branch exploration at high-uncertainty decision points, then consolidates test suites across trajectories and uses multi-model voting to select the best patch. Trigger phrases: "use EGSS to fix this bug", "entropy-guided scaling", "stepwise scaling for this task", "try multiple approaches with EGSS", "scale test-time compute for this fix", "use adaptive branching to solve this".
    14 repo stars
  43. ▌
    Evolve Evolutionary Search LLM Based · ndpvt-web
    Evolutionary search framework for LLM-driven Verilog/RTL generation and PPA optimization. Uses MCTS for functional correctness and Idea-Guided Refinement for optimization, with structured testbench generation for rapid feedback. Triggers: 'generate Verilog module', 'optimize RTL design', 'evolutionary search for hardware', 'MCTS Verilog generation', 'reduce PPA for circuit', 'fix Verilog functional correctness'.
    14 repo stars
  44. ▌
    Following Dragons Code Review Guided · ndpvt-web
    Extract security-relevant signals from code review comments and translate them into fuzzer-guiding annotations using the EyeQ pipeline. Use when the user says 'guide fuzzing from code reviews', 'find dragons in review comments', 'annotate code for fuzzing', 'review-guided fuzzing', 'extract security signals from PRs', or 'instrument code for AFL++ from review discussions'.
    14 repo stars
  45. ▌
    From Detection Prevention Explaining · ndpvt-web
    Proactively identify security-critical code regions and generate prevention-oriented explanations before vulnerabilities are introduced. Use when: 'review this code for security-critical areas', 'explain security risks in my methods', 'find security-sensitive code before bugs happen', 'proactive security review of my codebase', 'highlight authentication and data access risks', 'prevent vulnerabilities in this module'.
    14 repo stars
  46. ▌
    From Features Actions Explainability · ndpvt-web
    Diagnose and explain failures in agentic AI systems using trace-based rubric evaluation, bridging static feature attribution (SHAP/LIME) with trajectory-level diagnostics. Use when: 'debug why my agent failed', 'explain agent behavior', 'evaluate agent traces', 'add explainability to my agent pipeline', 'diagnose agentic failures', 'trace-based agent analysis'.
    14 repo stars
  47. ▌
    Gradingattack Attacking Short Answer · ndpvt-web
    Audit LLM-based automatic short answer grading (ASAG) systems for adversarial vulnerabilities using token-level and prompt-level attack strategies from the GradingAttack framework. Triggers: 'test grading robustness', 'adversarial attack on grading', 'audit LLM grader', 'red-team answer grading', 'ASAG vulnerability assessment', 'grading fairness attack'
    14 repo stars
  48. ▌
    Greprag Empirical Study Optimization · ndpvt-web
    Lightweight, index-free repository-level code retrieval using ripgrep for context-aware code completion. Uses LLM-generated ripgrep commands with identifier-weighted re-ranking and structure-aware deduplication. Trigger phrases: 'find relevant code across the repo', 'retrieve cross-file context for completion', 'grep the codebase for related code', 'use ripgrep to find dependencies', 'get repo context for this function', 'search the project for related definitions'
    14 repo stars
  49. ▌
    Hugrag Hierarchical Causal Knowledge · ndpvt-web
    Build hierarchical causal knowledge graphs for RAG pipelines that suppress spurious correlations and enable cross-document causal reasoning. Use when: 'build a causal knowledge graph from my documents', 'add causal reasoning to my RAG pipeline', 'set up graph-based RAG with causal filtering', 'create a hierarchical knowledge graph for retrieval', 'implement causal gating for my knowledge base', 'improve RAG with structured causal retrieval'.
    14 repo stars
  50. ▌
    Loca Bench Benchmarking Agents Under · ndpvt-web
    Apply context management strategies from LOCA-bench to prevent context rot in long-running agent tasks. Implements programmatic tool calling, tool-result clearing, thinking-block clearing, context awareness, and memory tools to maintain agent accuracy as context grows. Use when: 'my agent loses track after many steps', 'context window filling up', 'agent stops exploring midway', 'handle long context in agentic workflow', 'prevent context rot', 'manage growing conversation context'.
    14 repo stars
  51. ▌
    Medbeads Agent Native Immutable Data · ndpvt-web
    Build immutable, agent-native medical data pipelines using Merkle DAG structures (MedBeads pattern). Converts mutable EMR/FHIR records into cryptographically-linked, causally-ordered bead graphs that LLMs can traverse deterministically instead of relying on probabilistic RAG. Use when: 'build an immutable clinical data store', 'convert FHIR to a causal DAG', 'create tamper-evident medical records', 'agent-native healthcare data pipeline', 'deterministic context retrieval for medical AI', 'MedBeads implementation'.
    14 repo stars
  52. ▌
    Papersearchqa Learning Search Reason · ndpvt-web
    Build iterative search-and-reason agents for scientific literature QA. Uses the PaperSearchQA pattern: interleaved thinking, querying, and verification loops over document corpora. Trigger phrases: 'search scientific papers', 'build a paper search agent', 'scientific QA pipeline', 'iterative retrieval agent', 'RLVR search agent', 'reason over research papers'
    14 repo stars
  53. ▌
    Parse Open Domain Reasoning Question · ndpvt-web
    Build and evaluate reasoning-focused QA systems for low-resource languages using the PARSE methodology: structured prompting strategies (CoT for Boolean/multiple-choice, few-shot for factoid), multi-stage quality filtering, and language-native prompt design. Use when: 'build a Persian QA benchmark', 'evaluate reasoning in low-resource language', 'create multilingual QA dataset', 'optimize prompting strategy for question types', 'generate validated QA pairs with difficulty levels', 'fine-tune model for Persian question answering'.
    14 repo stars
  54. ▌
    Prism Xr Empowering Privacy Aware Xr · ndpvt-web
    Build privacy-aware pipelines that filter sensitive content from visual frames before sending to cloud AI models, using edge preprocessing with object detection (YOLO), text-based scene description, selective cropping, and structured MLLM interaction. Triggers: 'privacy-aware XR pipeline', 'filter sensitive data from camera frames', 'edge preprocessing before cloud AI', 'PRISM-XR privacy pipeline', 'sanitize visual input for LLM', 'multi-user XR collaboration with privacy'
    14 repo stars
  55. ▌
    Proact Agentic Lookahead Interactive · ndpvt-web
    Apply ProAct-style lookahead reasoning to multi-step coding and planning tasks. Compresses search-tree exploration into concise causal reasoning chains so the agent considers future consequences before committing to actions. Use when: "plan ahead before refactoring", "think through consequences of this change", "what happens if I restructure this module", "lookahead before migrating", "simulate outcomes of this architecture decision", "evaluate trade-offs for this design choice".
    14 repo stars
  56. ▌
    Reasoning Beyond Literal Cross Style · ndpvt-web
    Detect and interpret figurative language (sarcasm, humor, offense, metaphor) in multimodal image-text content using a structured five-step reasoning chain. Triggers: 'detect sarcasm in this meme', 'is this image sarcastic', 'analyze figurative meaning', 'interpret this humor/metaphor', 'classify meme sentiment', 'explain why this meme is funny/offensive'
    14 repo stars
  57. ▌
    Refuge Feature Generation Prediction · ndpvt-web
    Automated feature engineering for prediction tasks on relational databases using a multi-agent LLM pipeline. Generates, filters, and validates SQL/pandas features from multi-table schemas. Use when: 'generate features from my database for prediction', 'engineer features across related tables', 'automate feature generation for classification/regression on relational data', 'build predictive features from a star schema', 'improve model accuracy with relational feature engineering', 'create aggregation features from joined tables'.
    14 repo stars
  58. ▌
    Regular Variational Latent Reasoning · ndpvt-web
    Compress verbose chain-of-thought reasoning into compact latent state representations guided by rendered visual summaries, based on the ReGuLaR paper. Use when: 'compress my reasoning', 'latent reasoning over this problem', 'reduce CoT verbosity', 'efficient multi-step reasoning', 'render-guided reasoning compression', 'reason with fewer tokens'.
    14 repo stars
  59. ▌
    Ruleflow Generating Reusable Program · ndpvt-web
    Optimize Pandas code by discovering per-program improvements, generalizing them into reusable rewrite rules, and applying those rules as a lightweight compiler pass. Use when the user says 'optimize this Pandas code', 'speed up my DataFrame operations', 'rewrite my notebook for performance', 'apply Pandas rewrite rules', 'find slow Pandas patterns', or 'generate optimization rules for my Python data pipeline'.
    14 repo stars
  60. ▌
    Rulesmith Multi Agent Automated Game · ndpvt-web
    Automated game balancing using multi-agent LLM self-play coupled with Bayesian optimization. Use when the user asks to 'balance a game', 'tune game parameters', 'optimize game rules', 'automate playtesting', 'build a game balancing pipeline', or 'use LLM agents for game simulation'.
    14 repo stars
  61. ▌
    State Transition Framework Reasoning · ndpvt-web
    Applies a state-transition reasoning framework that models multi-step reasoning as an evolving state, compressing historical reasoning into a compact state representation and correcting noisy intermediate steps via momentum-based smoothing. Improves efficiency and accuracy on long reasoning chains. Trigger phrases: "reason step by step with state tracking", "use state-transition reasoning", "solve this with efficient chain-of-thought", "break this into reasoning states", "apply state-based reasoning", "reason efficiently without overthinking"
    14 repo stars
  62. ▌
    Svrepair Structured Visual Reasoning · ndpvt-web
    Fix bugs using structured visual reasoning -- converts screenshots, control-flow graphs, and UI artifacts into semantic scene graphs for precise fault localization and patch generation. Use when: 'fix this UI bug from the screenshot', 'the layout is broken, here is what it looks like', 'debug this using the control-flow graph', 'repair the rendering issue shown in this image', 'use the screenshot to find what went wrong', 'convert this visual bug report into a code fix'.
    14 repo stars
  63. ▌
    Textual Equilibrium Propagation Deep · ndpvt-web
    Optimize deep multi-step AI pipelines using Textual Equilibrium Propagation (TEP) — a two-phase local-then-nudge strategy that avoids gradient explosion/vanishing in long prompt chains. Use when: "optimize my multi-agent pipeline", "fix my prompt chain that degrades at depth", "improve my compound AI system", "TEP optimization", "local critic refinement for agents", "scale my LLM workflow without losing quality".
    14 repo stars
  64. ▌
    Usage Effects Requirements AI Coding · ndpvt-web
    Optimize AI coding assistant interactions using empirical enterprise findings on usage patterns, productivity factors, and quality requirements. Use when: 'help me get more out of Copilot', 'review my AI-assisted workflow', 'improve AI code generation quality', 'audit AI coding assistant effectiveness', 'optimize prompts for code generation', 'set up AI coding assistant practices for my team'.
    14 repo stars
  65. ▌
    Wdscaling Parallel Tool Calling Deep · ndpvt-web
    Scale deep research tasks by issuing parallel tool calls (width) alongside sequential reasoning (depth), following the Wide & Deep agent pattern. Use when: 'research this topic thoroughly', 'find information about X from multiple sources', 'deep dive into this question', 'investigate this from multiple angles', 'do a comprehensive search on X', 'compare sources on this claim'.
    14 repo stars
  66. ▌
    When Agents Misremember Collectively · ndpvt-web
    Detect, measure, and defend against collective false-memory propagation (the Mandela Effect) in LLM multi-agent systems. Use when: 'harden multi-agent pipeline against misinformation', 'audit agent consensus for false beliefs', 'add cognitive anchoring to agent prompts', 'defend agents from social influence bias', 'reduce reality shift in collaborative agents', 'mitigate groupthink in LLM swarms'.
    14 repo stars
  67. ▌
    Yunque Deepresearch Technical Report · ndpvt-web
    Hierarchical multi-agent deep research framework with dynamic context management and supervisor-based error recovery. Implements Yunque DeepResearch's orchestration pattern: decompose complex research questions into subtasks, route them to specialized tools/sub-agents, fold completed results into semantic summaries, and detect/recover from failures automatically. Use when: 'deep research on this topic', 'investigate this question thoroughly', 'multi-step research task', 'find and synthesize information from multiple sources', 'research agent for complex questions', 'autonomous research workflow'.
    14 repo stars
  68. ▌
    Zero Shot Product Attribute Labeling · ndpvt-web
    Extract and classify product attributes from images using Vision-Language Models with structured prompts and a three-tier evaluation framework. Handles conditional attributes (e.g., fabric type only when garment is visible) by separating applicability detection from classification. Triggers: 'extract product attributes from images', 'label fashion attributes zero-shot', 'classify clothing attributes with VLM', 'build product catalog enrichment pipeline', 'detect garment properties from photos', 'evaluate attribute prediction accuracy'.
    14 repo stars
  69. ▌
    Aiano Enhancing Information Retrieval · ndpvt-web
    Build AI-augmented annotation pipelines for creating high-quality information retrieval and QA datasets. Combines LLM-generated suggestions (questions, passage relevance scores, answer spans) with human review workflows to accelerate dataset creation. Use when: 'build an annotation pipeline for IR data', 'create a QA dataset from documents', 'annotate passages for retrieval training', 'generate question-answer pairs from a corpus', 'build a human-in-the-loop labeling tool for search', 'set up AI-assisted relevance annotation'.
    14 repo stars
  70. ▌
    Automated Rubrics Reliable Evaluation · ndpvt-web
    Generate fine-grained evaluation rubrics for medical dialogue systems using a retrieval-augmented multi-agent pipeline. Decomposes medical evidence into atomic facts, synthesizes them with interaction constraints, and produces weighted, auditable rubrics. Use when: 'evaluate medical chatbot responses', 'generate rubrics for clinical QA', 'build a medical LLM evaluation pipeline', 'score health dialogue quality', 'create automated clinical evaluation criteria', 'refine medical AI responses with rubric feedback'.
    14 repo stars
  71. ▌
    Bioace Automated Framework Biomedical · ndpvt-web
    Evaluate biomedical QA outputs using the BioACE nugget-based framework — assess answer completeness, correctness, precision, recall, and citation quality against ground-truth nuggets. Trigger phrases: - "evaluate biomedical answers" - "check citation quality for medical QA" - "nugget-based evaluation of RAG output" - "assess completeness and correctness of biomedical text" - "verify biomedical citations with NLI" - "score biomedical question answering output"
    14 repo stars
  72. ▌
    Capture Flags Family Based Evaluation · ndpvt-web
    Generate semantics-preserving variants of Python CTF challenges to stress-test agentic LLM robustness. Applies the Evolve-CTF methodology: identifier renaming, dead code insertion, composite transforms, and obfuscation to create challenge families that share a single exploit but vary in surface-level code. Trigger phrases: - "generate CTF variants" - "obfuscate this challenge" - "create a challenge family" - "test agent robustness on CTF" - "semantics-preserving transformation" - "evolve this CTF challenge"
    14 repo stars
  73. ▌
    Cvedrl Code Verifier Difficulty Aware · ndpvt-web
    Generate difficulty-aware unit tests that verify LLM-generated code using branch coverage analysis, complexity-weighted rewards, and majority voting selection. Use when asked to 'verify generated code', 'write tests for hard branches', 'difficulty-aware testing', 'rank code solutions with tests', 'generate verification tests', or 'test code with branch coverage'.
    14 repo stars
  74. ▌
    Dr Kernel Reinforcement Learning Done · ndpvt-web
    Write high-performance Triton GPU kernels using Dr. Kernel's multi-turn refinement strategy: profile-guided optimization, reward hacking prevention, and iterative kernel fusion. Use when asked to 'write a Triton kernel', 'optimize GPU code', 'fuse CUDA operations', 'speed up a PyTorch operation with Triton', 'generate a fast kernel for [operation]', or 'profile and optimize this kernel'.
    14 repo stars
  75. ▌
    Draincode Stealthy Energy Consumption · ndpvt-web
    Evaluate and defend RAG-based code generation systems against energy-drain attacks that poison retrieval contexts to inflate LLM output length, latency, and GPU energy consumption. Use when: 'audit my RAG pipeline for energy attacks', 'test code retrieval poisoning resilience', 'detect adversarial triggers in retrieved code', 'harden my code generation system against context poisoning', 'benchmark energy cost of retrieval-augmented code generation', 'simulate DrainCode-style attacks on my pipeline'.
    14 repo stars
  76. ▌
    Ecco Evidence Driven Causal Reasoning · ndpvt-web
    Apply evidence-driven causal reasoning to compiler optimization pass selection and ordering. Uses the ECCO framework: analyze static code features, build causal explanations linking features to pass effectiveness, then guide search-based optimization with LLM-generated optimization intents. Triggers: "optimize compiler passes", "find best LLVM pass order", "reduce execution cycles for this code", "causal compiler optimization", "why does this optimization pass help", "tune LLVM pass sequence for performance".
    14 repo stars
  77. ▌
    Embodied Task Planning Graph Informed · ndpvt-web
    Structure long-horizon task planning using graph-based memory and bounded lookahead. Use when asked to: 'plan a multi-step agent workflow', 'build a task planner for a robot or agent', 'decompose complex goals into grounded actions', 'implement graph-based planning memory', 'create an agent that learns from past executions', 'build a planning system that avoids hallucinated actions'.
    14 repo stars
  78. ▌
    Fraudshield Knowledge Graph Empowered · ndpvt-web
    Detect and defend against fraudulent content in LLM inputs using knowledge-graph-augmented analysis. Builds a fraud tactic-keyword bipartite graph, scores associations by confidence, prunes ambiguities, and augments prompts with XML-tagged keywords plus evidence rationales. Use when: 'check this email for fraud', 'is this job posting a scam', 'analyze this contract for manipulation', 'detect phishing in this message', 'flag suspicious text in this document', 'add fraud detection to my LLM pipeline'.
    14 repo stars
  79. ▌
    Helios Hierarchical Graph Abstraction · ndpvt-web
    Structure-aware binary decompilation using hierarchical control-flow graph abstraction for LLMs. Converts binary programs into compilable, semantically faithful C code by encoding CFG structure (basic blocks, successors, loops, conditionals) as a hierarchical text prompt alongside raw decompiler output, with optional compiler-in-the-loop feedback. Trigger phrases: 'decompile this binary', 'reverse engineer this executable', 'convert assembly to C', 'decompile with control flow', 'structure-aware decompilation', 'recompilable decompilation'
    14 repo stars
  80. ▌
    Large Model Powered Evolutionary Code · ndpvt-web
    Iteratively optimize code performance using LLM-driven evolutionary search on a phylogenetic tree. Applies PhyloEvolve-style mutation, crossover, elite trajectory pooling, and multi-branch exploration to improve runtime, memory, or correctness of algorithms. Use when: 'optimize this CUDA kernel', 'speed up this numerical solver', 'evolve better implementations of this algorithm', 'iteratively improve this function performance', 'find a faster version of this code', 'reduce memory usage of this computation'.
    14 repo stars
  81. ▌
    Learning Compose Cross Domain Agentic · ndpvt-web
    Generate cross-domain agentic workflows using decompose-recompose-decide composition over reusable capability bases. Use when asked to: 'design a multi-step agent workflow', 'create a pipeline that works across different task types', 'build a reusable agentic workflow', 'compose agent operators for a new domain', 'generate a single-pass workflow for complex tasks', 'orchestrate reasoning-verification-repair agents'.
    14 repo stars
  82. ▌
    Learning Reason Faithfully Step Level · ndpvt-web
    Apply FaithRL's step-level faithfulness verification to multi-step reasoning tasks. Decomposes reasoning into individually verified steps, penalizes unsupported claims, and preserves valid partial derivations. Use when: 'verify each reasoning step', 'reduce hallucination in chain-of-thought', 'faithful step-by-step reasoning', 'audit my reasoning chain', 'step-level credit assignment', 'penalize unsupported reasoning steps'.
    14 repo stars
  83. ▌
    Linguistagent A Reflective Multimodel · ndpvt-web
    Implements a reflective dual-agent (Annotator + Reviewer) workflow for automated linguistic annotation tasks such as metaphor identification, sentiment labeling, named entity recognition, and other sequence-labeling problems. The Annotator marks spans in text using XML tags and provides reasoning; the Reviewer critiques the annotations against a codebook, catching false positives and missed instances, then feeds corrections back for self-improvement. Trigger phrases: "annotate this text for metaphors", "dual-agent annotation pipeline", "reflective annotation workflow", "linguistic annotation with LLM review", "peer-review annotation system", "automated text labeling with self-correction"
    14 repo stars
  84. ▌
    Patch To Poc Systematic Study Agentic · ndpvt-web
    Agentic kernel vulnerability reproduction from security patches. Implements the K-Repro methodology: controlled code browsing, hypothesis-driven root cause analysis, iterative PoC generation, VM-based testing, and GDB debugging for Linux kernel N-day reproduction. Trigger phrases: "reproduce this kernel vulnerability", "generate a PoC from this patch", "analyze this security patch for exploitability", "kernel bug reproduction from commit", "patch-to-poc analysis", "N-day vulnerability reproduction"
    14 repo stars
  85. ▌
    Research Multi Stage Machine Learning · ndpvt-web
    Build multi-stage search pipelines that separate recall from precision for discovering datasets, documents, or resources from heterogeneous metadata catalogs. Uses intent classification, hybrid retrieval (BM25 + embeddings + abbreviation expansion), and LLM reranking. Trigger phrases: 'build a dataset discovery pipeline', 'multi-stage search system', 'find relevant datasets for my research goal', 'scientific data search', 'hybrid retrieval with reranking', 'intent-aware search pipeline'
    14 repo stars
  86. ▌
    Sparc Separating Perception Reasoning · ndpvt-web
    Decouple visual perception from reasoning when building VLM pipelines, image analysis agents, or multi-modal workflows. Implements the SPARC two-stage pattern: first localize question-relevant regions via visual search, then condition reasoning only on those regions. Dramatically reduces token budgets and improves accuracy on visually complex tasks. Trigger phrases: - "Build an image QA pipeline that scales at test time" - "Analyze this high-resolution image efficiently with a VLM" - "Separate perception from reasoning in my vision pipeline" - "Reduce token cost for visual reasoning tasks" - "Implement a two-stage visual search and answer system" - "Scale VLM inference with asymmetric compute allocation"
    14 repo stars
  87. ▌
    Supporting Software Engineering Tasks · ndpvt-web
    Generate test scenarios from requirements and retrieve/analyze software engineering documents using a supervisor-worker star topology of specialized agents. Use when: 'generate test cases from requirements', 'search project documents', 'summarize this specification', 'track changes across document versions', 'answer questions about our design docs', 'create test scenarios from this feature spec'.
    14 repo stars
  88. ▌
    Swe Replay Test Time Scaling Software · ndpvt-web
    Efficient test-time scaling for software engineering agents using trajectory recycling and explore-exploit branching (SWE-Replay). Use when: 'scale up my agent to solve this hard bug', 'retry this task more efficiently', 'run multiple attempts on this SWE task', 'replay and branch from a previous attempt', 'use test-time compute to improve solve rate', 'efficiently sample multiple solutions for this issue'.
    14 repo stars
  89. ▌
    Towards AI Evaluation Domain Specific · ndpvt-web
    Build and evaluate domain-specific RAG systems with iterative user-feedback refinement, source grounding, and structured evaluation pipelines. Use when: 'build a RAG system for [domain] documents', 'evaluate my RAG pipeline quality', 'add user feedback to my retrieval system', 'improve RAG answers with source grounding', 'set up domain-specific document retrieval', 'create an evaluation framework for my RAG app'.
    14 repo stars
  90. ▌
    Understanding Agent Scaling LLM Based · ndpvt-web
    Design diversity-aware multi-agent systems that maximize performance with fewer agents. Uses information-theoretic K* effective channel analysis to replace brute-force agent scaling with principled heterogeneous configurations. Trigger phrases: 'optimize my agent swarm', 'why are my agents not improving with scale', 'design a diverse multi-agent system', 'K-star analysis for agents', 'reduce agent count without losing accuracy', 'heterogeneous agent configuration'
    14 repo stars
  91. ▌
    Unicog Uncovering Cognitive Abilities · ndpvt-web
    Analyze and diagnose LLM reasoning through latent cognitive ability decomposition inspired by the UniCog framework. Decomposes reasoning traces into sparse cognitive dimensions to identify failure modes, rank candidate solutions, and improve multi-step reasoning accuracy. Use when: 'diagnose why the model reasoning failed', 'analyze cognitive load of these solutions', 'rank candidate answers by reasoning quality', 'find the reasoning failure in this chain of thought', 'decompose this reasoning into cognitive abilities', 'prioritize solutions using activation analysis'.
    14 repo stars
  92. ▌
    Adaptive Confidence Gating Multi Agent · ndpvt-web
    Multi-agent code generation using structured debate with adaptive confidence gating. Three specialized agents (User/Product, Technical, QA) debate before generating code, with a 95% confidence threshold to skip debate on simple tasks and a reviewer-guided debugging loop for post-generation refinement. Use when: 'generate code with multi-agent debate', 'use DebateCoder approach', 'code generation with confidence gating', 'multi-agent collaborative coding', 'debate-driven code synthesis', 'adaptive confidence code generation'.
    14 repo stars
  93. ▌
    Adareasoner Dynamic Tool Orchestration · ndpvt-web
    Adaptive multi-step tool orchestration for complex reasoning tasks. Dynamically selects, sequences, and composes tools based on task context and intermediate results rather than fixed pipelines. Use when: 'orchestrate tools for this task', 'figure out which tools to use', 'multi-step reasoning with tools', 'adaptive tool pipeline', 'dynamic tool selection', 'chain tools together intelligently'.
    14 repo stars
  94. ▌
    Aero Autonomous Evolutionary Reasoning · ndpvt-web
    Apply the AERO dual-loop self-evolution framework to iteratively improve reasoning on complex tasks. Uses entropy-based difficulty calibration, counterfactual verification, and staggered role refinement to solve hard problems without external oracles. Triggers: 'reason through this step by step with self-correction', 'solve this hard problem autonomously', 'verify your reasoning with counterfactuals', 'self-critique and improve your answer', 'use dual-loop reasoning on this', 'iteratively refine your solution'
    14 repo stars
  95. ▌
    Agentxray White Boxing Agentic Systems · ndpvt-web
    Reverse-engineer black-box agentic systems into editable, interpretable workflows using search-based reconstruction. Use when the user says 'reconstruct this agent workflow', 'reverse-engineer this pipeline', 'white-box this agentic system', 'explain what this agent chain is doing', 'approximate this black-box agent', or 'build an interpretable surrogate for this system'.
    14 repo stars
  96. ▌
    Automated Structural Testing LLM Based · ndpvt-web
    Write structural tests for LLM-based agents using trace-based assertions, mocked LLM responses, and the test automation pyramid. Use when the user says 'test my agent', 'write agent tests', 'mock LLM responses', 'add regression tests for my agent', 'structural testing for agents', or 'trace-based assertions'.
    14 repo stars
  97. ▌
    Autonomous Data Processing Meta Agents · ndpvt-web
    Build self-managing data processing pipelines using hierarchical meta-agent orchestration. Decomposes complex data tasks into multi-phase plans, instantiates specialized ground-level agents (Reader, Profiler, Transformer, Validator, Joiner, etc.), and iteratively refines execution through progressive sampling and monitoring loops. Triggers: 'build a data pipeline', 'process and clean this dataset', 'create an ETL pipeline with agents', 'orchestrate data processing agents', 'autonomous data pipeline', 'meta-agent data processing'.
    14 repo stars
  98. ▌
    Benchmarking Pairwise Causal Discovery · ndpvt-web
    Detect and extract pairwise causal relationships from text using structured prompting strategies (zero-shot, CoT, few-shot ICL, Least-to-Most, ReAct). Built on the unified evaluation framework from Anuyah et al. (2026) covering 12 datasets across biomedical and multi-domain contexts. Use when the user asks to "find causal relationships in text", "extract cause and effect from sentences", "detect causation vs correlation", "build a causal extraction pipeline", "benchmark causal reasoning", or "annotate causal links in a corpus".
    14 repo stars
  99. ▌
    Beyond Translation Cross Cultural Meme · ndpvt-web
    Cross-cultural meme transcreation using a three-stage hybrid pipeline (cultural analysis, visual generation, assembly) that preserves humor and communicative intent while adapting culture-specific references between languages. Triggers: 'transcreate this meme', 'adapt meme for Chinese audience', 'convert meme to US culture', 'cross-cultural meme adaptation', 'localize this meme for another culture', 'meme cultural translation'
    14 repo stars
  100. ▌
    Cam Causality Based Analysis Framework · ndpvt-web
    Analyze and optimize multi-agent code generation pipelines using causality-based importance ranking of intermediate features. Identifies which pipeline stages matter most, enables targeted failure repair, token-efficient pruning, and hybrid LLM backend assignment. Triggers: 'analyze my multi-agent pipeline', 'optimize agent code generation', 'which pipeline stages matter most', 'reduce token usage in my agent system', 'fix failing multi-agent code generation', 'assign LLMs to pipeline stages'
    14 repo stars