Skill UI

A practical decision framework for determining whether to use regular expressions or Large Language Models (LLMs) when parsing structured text (e.g., forms, quizzes, invoices). It recommends starting with deterministic, low-cost regex for 95-98% of structured data, and only invoking expensive LLM calls for identifying and validating ambiguous edge cases, optimizing both cost and accuracy.

This critical skill serves as a gate between completed experiments and final research conclusions. It systematically evaluates collected experimental results (e.g., W&B metrics, logs) against predefined research claims. It performs deterministic evidence pre-checks to verify if cited data points exist, and utilizes an advanced AI model (Codex) to judge whether the results support, contradict, or lack evidence for the intended claims, guiding the next action (pivot, supplement, or confirm).

This internal agent skill processes raw X/Twitter posts into structured bug candidates. It systematically performs parsing, content deduplication, PII redaction, reliability scoring, and classification. Finally, it clusters related reports using deterministic signatures, providing a highly organized output for efficient bug triage and analysis from large-scale social media streams.

This advanced skill acts as a primary research orchestrator. It deterministically analyzes user queries, routing them to specialized modules (e.g., patent landscape, academic literature review, funding intelligence) or executing a comprehensive fallback search workflow. It is designed for complex, multi-faceted research requiring high rigor, structured planning, and verifiable citations.

Bilig WorkPaper provides a code-first runtime environment for complex spreadsheet business logic. It allows agents to interact with structured data using formulas and JSON workbooks via deterministic API calls (MCP/TypeScript). This is ideal for replacing unreliable UI automation in use cases like quote calculators, budget modeling, and data validation, ensuring calculation integrity through explicit read-back proof.

A powerful tool for deterministic and reproducible retrieval of evidence from specialized academic and policy corpora. It supports searching arXiv (preprints), PubMed/PMC (biomedical literature), and US policy documents. Users can check corpus freshness, retrieve stable metadata, and fetch full-document Markdown for high-reliability research tasks.

This utility serves as a stub experiment generator, primarily for demonstration purposes within meta-paper writing workflows. Given a specific topic phrase, it deterministically produces a consistent CSV file containing simulated experimental results. This function allows users to prototype and test complex paper structures and formats without needing access to real scientific data.

A practical decision framework for determining whether to use regular expressions or Large Language Models (LLMs) when parsing structured text (e.g., forms, quizzes, invoices). It recommends starting with deterministic, low-cost regex for 95-98% of structured data, and only invoking expensive LLM calls for identifying and validating ambiguous edge cases, optimizing both cost and accuracy.