Skill UI

This comprehensive guide details the rigorous methodology for designing, executing, and reporting experimental results in Information Retrieval (IR). It covers best practices for selecting appropriate test collections, choosing defensible metrics, performing proper statistical significance testing (e.g., paired t-tests), ensuring baseline fairness, and proactively addressing modern pitfalls like LLM contamination. Essential for publishing high-quality, auditable research in IR/NLP.

A comprehensive guide on structuring the 'Related Work' section for top-tier Information Retrieval (IR) conferences (like SIGIR). Learn to trace the lineage of your research through canonical history (e.g., BM25, ColBERT), mechanically articulate your technical 'delta' (novel contribution), and avoid common pitfalls like venue misattribution and insufficient contrast.

A comprehensive guide for ensuring the reproducibility of Information Retrieval (IR) research, essential for SIGIR submissions. It details best practices for pinning data collection versions, index settings, neural model hyperparameters, and managing non-determinism. It helps researchers document artifacts—not just promises—to validate reported gains.

This comprehensive guide assists researchers in determining the core contribution of their Information Retrieval (IR) work. It provides structured tests to decide whether the contribution lies in retrieval quality or the downstream system. Users learn how to select the appropriate SIGIR track (Full, Short, Resources, etc.) or route their work to neighboring venues like NLP/ML, RecSys, or CHIIR.

A comprehensive guide for authors aiming to publish in top-tier Information Retrieval (IR) conferences like SIGIR. It teaches the specific, quantitative 'native register' required, moving beyond general machine learning boilerplate. Covers structuring the first page arc, calibrating vocabulary, and building table-centric arguments under strict page limits.

This skill guides authors on positioning information retrieval (IR) work for the Text REtrieval Conference (TREC). It functions as a venue-selection and manuscript re-framing tool, ensuring the submission aligns with TREC's focus on shared evaluation tracks, test collections, and NIST-style benchmarking. It helps authors structure evidence, understand reviewer expectations, and distinguish the work from general CS conference submissions.

This guide provides a comprehensive framework for designing rigorous and defensible evaluation sections for machine learning and information retrieval papers, particularly for top-tier conferences like WSDM. It covers the four critical evidence quadrants—Effectiveness, Validity, Efficiency, and Robustness—and details necessary protocols, such as temporal splits, proper baseline selection (including heuristics), ablation studies, and hardware-aware performance reporting. Use this when auditing or writing the experimental section of a research paper.

This skill serves as a deterministic internal downloader for image URLs extracted from comprehensive web search results. It receives normalized media slots and search output via stdin, utilizing Python HTTP APIs to download direct image links. The process includes validating MIME types and saving files accurately into the configured local project directory, ensuring reliable asset retrieval without relying on sandboxed shell tools.

A specialized downstream skill for analyzing security vulnerabilities in complex agentic systems. This tool is designed for red teaming and penetration testing, focusing on detecting prompt injection, retrieval poisoning, memory contamination, planner drift, and tool boundary abuse within multi-step LLM chains. Use when proving how untrusted inputs compromise system integrity.

This skill implements an advanced reference pattern for integrating with Monitoring Control Platform (MCP) servers. It ensures accurate data retrieval by first reading a domain-specific 'gotchas' reference file, mitigating query failures caused by complex server behaviors, pagination limits, or subtle API quirks. It analyzes issue data, groups errors by root cause, and generates a structured, actionable markdown report for improved system health visibility.

This comprehensive skill allows users to search the web using the DuckDuckGo API, retrieving diverse content including text results, news articles, images, and videos. It supports advanced filtering options such as time range, region, content type (e.g., news, images), and specific media characteristics (e.g., image size, video duration). Ideal for research, fact-checking, academic study, and gathering current web resources.

This utility provides a straightforward mechanism for reading the contents of specified files and returning the data. It is ideal for quick file inspection, data validation, or integrating basic file content retrieval into automated scripts and workflows.