Skill UI

This skill implements a Manus-style, file-based planning system, providing a persistent 'working memory' on disk. It helps users structure, track, and manage complex, multi-step projects, intensive research tasks, or detailed workflows by creating and maintaining dedicated markdown files (`task_plan.md`, `progress.md`, `findings.md`). Use this when tackling large tasks requiring detailed breakdown or multiple tool calls.

Researcher persona keeps references, notes, and teams coordinated by using Drive folders, Docs summaries, Sheets logs, and Gmail-aware sharing plus workflow announcements for structured collaboration.

PIP kicks in when tasks fail repeatedly or the assistant becomes passive, layering Western big-tech performance rhetoric, a systematic debugging methodology, and proactivity enforcement so every code, ops, research, deployment, or writing task is owned, verified, and driven to completion.

A robust, file-based system designed for managing and tracking complex, multi-step projects and research tasks. It uses persistent Markdown files (task_plan.md, findings.md, progress.md) as external working memory to ensure context continuity, facilitate session recovery, and maintain a structured, auditable record of project progress, regardless of the number of tool interactions.

Generates professional conference presentation slides from a compiled research paper. This skill produces dual outputs: a polished Beamer LaTeX PDF for high-quality typesetting, and an editable PPTX for flexible adjustments. It supports multiple talk types (Poster, Spotlight, Oral, Invited) and includes detailed guidelines for speaker notes and talk scripting, making it essential for academic researchers preparing for major ML and systems conferences.

This guide details advanced techniques for searching cricket matches by leveraging precise numerical scoring data. Instead of broad terms, users should utilize specific details—such as team totals (e.g., 168/9), runs off balls, wickets taken, or over details—combined with the match type and year to create highly targeted search queries. This method significantly improves search accuracy and efficiency when researching sports statistics.

This technique guides users on decomposing multi-constraint questions, especially in the domain of football trivia or historical research. Instead of searching all criteria simultaneously, it instructs users to rank constraints by distinctiveness (e.g., specific match results first, then historical milestones). By performing incremental, cascading searches, the system dramatically narrows down the result set, leading to more accurate information retrieval.

A guide for refining search queries when researching Indian or South Asian actors. It emphasizes the necessity of including domain keywords such as "Bollywood," "Hindi film," or specific Indian award names (e.g., "Filmfare Award") to filter out irrelevant Western search results and drastically improve search precision.

This skill transforms an LLM into a disciplined wiki maintainer, enabling the creation of a persistent, compounding personal knowledge base (Second Brain) within Obsidian. Unlike standard RAG workflows, which retrieve and synthesize information at query time, this system actively ingests sources, automatically updates entity pages, maintains cross-references, and continually synthesizes knowledge. Ideal for deep research, personal journaling, or building comprehensive business wikis over time.

This technique guides users through a rigorous two-phase research strategy. First, conduct a broad search to identify potential papers or authors. Crucially, the system instructs the user to use the extracted "author full name" as a highly specific anchor in subsequent queries. Combining the author's name with details like journal, institution, year, or biographical information ensures the most precise and reliable results, significantly narrowing down the search scope and achieving deep academic insights.

This tool functions as a post-task epilogue, designed to audit and record the entire research journey. It scans the conversation history at the end of a coding or research session to extract critical research artifacts, including decisions, experiments, dead ends, claims, heuristics, and major pivots. It systematically writes this provenance data into the structured `ara/` directory, ensuring a faithful and auditable trace of how the project truly evolved.

The Exa Web Toolkit provides advanced capabilities for high-quality web search and content extraction. It excels at handling scientific, technical, and academic queries, allowing users to filter results by specialized categories like 'research paper' and specific academic domains. Beyond search, it can fetch and extract content from multiple URLs, including articles and academic PDFs, making it essential for deep scholarly research and current information gathering.