wjgoarxiv

Installs and uses Antigravity CLI ASW skills, hooks, diagnostics, and swarm workflow guidance.

Remove AI-looking clutter and temporary artifacts without changing behavior.

Hypothesis-driven Antigravity Swarm debugging for crashes, hangs, wrong output, and runtime drift.

Antigravity Swarm Loop executes RED to GREEN to real-surface QA with cleanup receipts.

Antigravity Swarm Plan creates a decision-complete plan before large or ambiguous work.

Strict Antigravity implementation discipline for Python, TypeScript, JavaScript, Go, and Rust work.

Behavior-preserving Antigravity Swarm refactoring with characterization, impact mapping, diagnostics, and real-surface verification.

Antigravity Swarm review skill for diff, evidence, hook, package, docs, installer, and real-surface QA inspection.

Autonomous research and experimentation toolkit with 10 commands. Core loop inspired by Karpathy's autoresearch — generalizes to any domain with mechanical evaluation, overnight persistence, and zero dependencies. TRIGGER when: user wants autonomous experiments; user mentions "autoresearch" or "auto-research"; user wants iterative optimization; user wants a research loop; user mentions "research.md"; user wants to iterate until some condition; user wants to optimize code, prompts, configs, or parameters iteratively; user invokes any /autoresearch:* subcommand. DO NOT TRIGGER when: user wants a one-shot answer; user wants manual step-by-step guidance; user just wants to read a single paper; user wants a simple web search.

Core autonomous research loop. Reads research.md, proposes hypotheses, runs experiments, evaluates results mechanically, keeps improvements, discards failures, and iterates until the target metric is achieved or the iteration budget is exhausted. TRIGGER when: user invokes "autoresearch" (no subcommand); research.md exists; user wants the 5-stage loop; user wants iterative optimization overnight.

Feedback-driven self-improvement protocol for autoresearch-skill. Converts failed runs, confusing transcripts, bad outputs, or user feedback into a bounded improvement plan, an eval scenario, and a patch checklist without executing the patch automatically. TRIGGER when: user says the skill failed, wants to improve autoresearch-skill, provides feedback on confusing behavior, asks to turn a failure into an eval, or invokes /autoresearch:learn. DO NOT TRIGGER when: the user wants to run an optimization loop (use /autoresearch), fix code errors (use /autoresearch:fix), or ship a finished artifact (use /autoresearch:ship).

Comprehensive PDF manipulation toolkit for extracting text and tables, creating new PDFs, merging/splitting documents, and handling forms. When the LLM (Claude, ChatGPT, Gemini, or others) needs to fill in a PDF form or programmatically process, generate, or analyze PDF documents at scale.

Scientific bug hunting using falsifiable hypotheses. Forms hypotheses, designs falsifying tests, eliminates candidates systematically, and logs the full investigation trail in a structured debug/ folder. TRIGGER when: user has a bug to investigate scientifically; user wants systematic root-cause analysis; user says "debug", "investigate", "root cause", "why is this failing"; user invokes /autoresearch:debug. DO NOT TRIGGER when: user wants to optimize a metric (use /autoresearch); user wants to fix a known error automatically (use /autoresearch:fix); user just wants a quick one-line answer about what a function does.

Iterative error-crusher loop that auto-stops at 0 errors. Cascade-aware: fixes dependency errors before their dependents. Refuses anti-patterns that hide errors instead of fixing them. TRIGGER when: user has errors or failures to fix iteratively; user asks to "fix all errors"; user has a failing test suite; user has compilation errors; user has linter errors; user wants systematic error elimination; user invokes /autoresearch:fix. DO NOT TRIGGER when: user wants a one-shot fix for a single obvious bug; user wants debugging guidance only; user wants code review without fixing.

7-step setup wizard that produces a complete, ready-to-run research.md without executing the research loop. Walks the user through goal, metric, search space, constraints, evaluator design, and baseline measurement, then writes the file. TRIGGER when: user wants to set up a research project; user wants to plan before running the loop; user says "plan my research"; user has a goal but no research.md; user invokes /autoresearch:plan. DO NOT TRIGGER when: research.md already exists and the user wants to run the loop; user wants a one-shot answer; user wants to debug, not optimize.

Multi-perspective deliberation engine. Gathers independent positions from diverse personas, runs cross-examination and rebuttal rounds, detects herd behavior, and synthesizes a neutral judge verdict with confidence levels. TRIGGER when: user wants multi-perspective prediction, forecasting, scenario analysis, decision analysis, "what will happen if", "should we", "predict the outcome of", structured devil's advocacy, or any question benefiting from adversarial deliberation.

Multi-agent research conference that spawns parallel autoresearchers with symposium sharing, adversarial peer review, and insight synthesis. Reads a conference.md, orchestrates N researchers in structured rounds, and produces a synthesized result combining the best findings. TRIGGER when: user mentions "autoconference" or "conference" with research context; user wants multiple researchers competing or collaborating; user wants parallel autoresearch with peer review; user mentions "conference.md"; user wants research synthesis from multiple approaches; user wants adversarial evaluation of research results. DO NOT TRIGGER when: user wants a single autoresearch loop (use autoresearch-skill); user wants a simple one-shot answer; user wants to read a single paper.

Core conference loop — spawns N parallel autoresearchers, runs 4-phase rounds (independent research, poster session, peer review, knowledge transfer), and produces synthesized results combining the best findings.

Post-conference analysis — compares researcher trajectories, identifies failure modes, extracts transferable insights, and produces an insight taxonomy. TRIGGER when: user wants to analyze conference results, understand what happened, extract learnings from a completed conference. DO NOT TRIGGER when: user wants to run a conference (use autoconference) or ship results (use ship).

Adversarial 2-researcher debate mode. One researcher argues Position A (pro), the other argues Position B (con). Structured rounds with an Opus judge. TRIGGER when: user wants a debate, adversarial evaluation, pro/con analysis, wants to test opposing viewpoints. DO NOT TRIGGER when: user wants collaborative multi-researcher conference (use autoconference).

8-step interactive wizard that produces a fully configured conference.md file. TRIGGER when: user wants to set up a new conference, plan a conference, create conference.md. DO NOT TRIGGER when: user wants to run an existing conference (use autoconference).

Resume an interrupted autoconference from its last checkpoint. Reads conference_events.jsonl to find last completed phase and resumes. TRIGGER when: user wants to resume/continue a conference, conference was interrupted. DO NOT TRIGGER when: user wants to start a new conference (use autoconference or plan).

8-phase pipeline to convert conference results into paper-ready output. Formats, verifies, and packages conference findings for publication. TRIGGER when: user wants to ship/publish/format conference results, create a paper/report/blog post from conference output. DO NOT TRIGGER when: user wants to run a conference (use autoconference).

Systematic multi-database literature survey. 3 researchers cover non-overlapping sources (databases, time periods, or methodologies) with citation-chain knowledge transfer. TRIGGER when: user wants a literature survey, systematic review, paper survey, wants to cover multiple databases or time periods. DO NOT TRIGGER when: user wants to optimize a metric (use autoconference) or read a single paper (use scientific-reading).

Use BEFORE implementing anything non-trivial — when a request is broad, multi-step, ambiguous, or risky and needs a decision-complete plan. Runs explore-first research, an interview on genuine unknowns, writes and self-reviews one plan, then stops at an explicit approval gate.

Activate when the user types `csw` (alone, or as a prefix like `csw <task>`), or asks to "run the loop" / work with full rigor. Runs the evidence-bound autonomous loop — bind a goal, plan if needed, then drive each criterion through test-first implementation, real manual QA, independent review, and paired cleanup until the completion oracle passes.

Use to run a rigorous multi-lane review before declaring work done — for any broad, risky, shared, security-sensitive, or release-facing change. Uses native scheduling for independent review lanes and gates on ALL of them passing with a single binding verdict.

Use to EXECUTE an approved plan with discipline. Resumes from durable .csw/ goal state, drives each task through test-first implementation, real manual QA, independent verification, and paired cleanup, and only finishes when the completion oracle passes. Bootstraps csw-plan first if no plan exists.

Use when work contains two or more INDEPENDENT subtasks that can run at once (parallel investigation, multi-file changes, fan-out research, multi-lane review). Uses native Copilot task/fleet scheduling and integrates worker results skeptically.

COMSOL Multiphysics toolkit for .mph model analysis, input preparation, mesh/geometry editing, and MCP connectivity. Use when (1) user provides a .mph file and wants to understand its contents, (2) preparing simulation inputs or parameter sweeps, (3) modifying .mph parameters without COMSOL, (4) generating Java API code from .mph models, (5) editing mphtxt mesh or STL geometry files (translate, rotate, scale, mirror), (6) converting STL to COMSOL mphtxt format, (7) checking COMSOL installation and MCP readiness, (8) setting up COMSOL MCP server for Claude control. Triggers on: .mph files, COMSOL model, analyze mph, parameter sweep, comsolbatch, mesh editing, STL editing, mphtxt, COMSOL MCP, simulation inputs.

Extracts transcripts from YouTube videos and produces structured digests with TL;DR, key takeaways, core assertions with timestamps, topic timeline, and notable quotes. Builds on markitdown's raw transcript extraction by adding LLM-driven analysis and structured output. TRIGGER when: user pastes a YouTube URL and wants a summary, digest, or analysis of the video content; user asks "what is this video about"; user wants key points or takeaways from a YouTube video; user wants to decide whether a video is worth watching; user asks to summarize a YouTube talk, lecture, podcast, or interview; user wants structured notes from a video; user asks to digest or break down a YouTube video; user mentions a YouTube link and asks for insights. DO NOT TRIGGER when: user just wants the raw transcript text without analysis (use markitdown); user wants to download the video file; user wants to convert a local video or audio file to text.

Convert the current working directory into a task-local LLM-maintained wiki. Use when the user wants to organize project knowledge, ingest files or URLs into a local markdown wiki, bootstrap a wiki structure in-place, maintain an existing mini wiki, or build a repo-scoped knowledge base instead of a giant global vault. Also use when scientific papers, PDFs, TeX bundles, bibliographies, or research corpora should be ingested into a maintained local wiki with provenance and citations, subject to extraction quality and available metadata. Trigger on phrases like "wiki this repo", "ingest raw/", "build a local wiki", "maintain this knowledge base", "turn this directory into an LLM wiki", or when repeated project knowledge needs to compound across sessions. DO NOT TRIGGER when: the user only wants a one-shot summary, a global PKM redesign, or generic markdown cleanup with no persistent wiki intent.

Award-winning frontend design skill producing hand-crafted, human-quality interfaces. Generates landing pages, dashboards, web apps, components, and posters with distinctive typography, bold color systems, intentional motion, and spatial compositions that look designed by an expensive agency — NOT generic AI output. Use this skill whenever the user wants a landing page, dashboard, web app, component, poster, or any frontend UI. Also use when user asks for "beautiful", "premium", "award-winning", "stunning", or "polished" frontend; wants HTML/CSS/React/Vue/Svelte UI; mentions "design" with frontend context; wants to build a website; says "make it look good" or "make it pretty"; asks for UI that doesn't look like AI generated it; mentions "anti-slop" or "no AI aesthetic"; wants to "beautify", "style", or "redesign" an existing UI; asks for a "modern", "professional", or "polished" website — even if they don't explicitly mention design quality. Make sure to use this skill for ANY frontend creation task. DO NOT T

Creates or updates a resolved HANDOFF.md continuation document that captures progress, findings, and remaining work so a fresh agent can continue without context loss. TRIGGER when: the current agent cannot complete a task and needs to pass it to a new agent, context window is nearly full, user explicitly requests a handoff, or the skill is invoked as bare `handoff`. DO NOT TRIGGER when: the task is normal ongoing work with no agent transition needed.

Converts bidirectionally between DOCX/PDF and Markdown, edits existing DOCX files programmatically, or prepares a journal-ready manuscript. TRIGGER when: user needs DOCX↔Markdown conversion, wants to create a styled Word document from Markdown, needs to batch-edit/modify an existing DOCX file, or wants to produce a publisher-styled manuscript (e.g. Elsevier, ACS, IEEE, Nature, or a Korean generic profile). DO NOT TRIGGER when: user only needs to convert non-DOCX formats (images, HTML, etc.) to Markdown — use markitdown instead.

Creates, edits, and analyzes spreadsheet files (.xlsx, .xlsm, .csv) with formulas, formatting, data analysis, and visualization. TRIGGER when: user needs to create or modify a spreadsheet, run formula-based calculations, analyze tabular data, or build financial models in Excel/CSV format. DO NOT TRIGGER when: user is working with PDF documents — use the pdf skill instead.

Ingests Outlook desktop mail from live Outlook profiles, .msg files, and .pst archives into a local workspace and SQLite index, then extracts summaries, tasks, deadlines, decisions, and follow-ups. Use this skill whenever the user wants Outlook mailbox analysis, weekly or monthly reporting from mail, local mail archiving, task extraction from Outlook messages, or Outlook-side actions such as moving mail or creating Outlook tasks/calendar items. Ask interactively for semantic ambiguity and high-impact execution.