PAI Skills Catalog

Comprehensive research, analysis, and content research system with 3 modes (quick/standard/extensive) and 242+ Fabric patterns.

YouTube research using yt-dlp. Search by keyword, analyze videos, get transcripts, find trending content.

Search and retrieve arXiv papers by topic, category, or ID, with AlphaXiv AI-generated overviews across cs.AI, cs.LG, cs.CL, and more.

240+ specialized patterns for content analysis, extraction, and transformation. Pattern-based content processing system.

Google NotebookLM integration. Create research notebooks, add sources, generate audio overviews, mind maps, and flashcards.

Web security assessment and penetration testing. Vulnerability scanning for web applications.

Adversarial triage of raw security findings. Verify, deduplicate, rank by exploitability, route to owner.

STRIDE-per-asset threat modeling and security remediation for PAI infrastructure. Ranked findings with executable remediations.

Review recently changed code for opportunities to improve reuse, quality, and efficiency. Post-change review workflow.

Agent evaluation framework based on Anthropic's best practices. Three grader types (code-based, model-based, human) with pass@k metrics.

Generate TypeScript CLIs with proper argument parsing, help text, and error handling.

Dynamic wisdom extraction that adapts sections to content. Extract insights from videos, podcasts, articles.

Document processing for DOCX, PDF, PPTX, XLSX. Extract, parse, and analyze structured documents.

Parse URLs, files, articles, transcripts, and PDFs into structured JSON. General content extraction.

Unified debate/council router. Routes to the right mechanism automatically for structured multi-perspective reasoning.

Extended thinking mode for creative problem solving and deep reasoning.

Evolutionary ideation engine with 9-phase loop-controlled multi-cycle generation. Novel candidate generation with provenance.

Adversarial analysis with 32 agents. Attack ideas, find counterarguments, stress test assumptions.

First principles analysis. Decompose to fundamentals, identify core assumptions, rebuild from ground up.

Infrastructure health probe. Checks Qdrant, Ollama, systemd services, backups, and disk space in one call.

Debug-first browser automation with always-on visibility. Console logs, network requests, and errors captured by default.

Search prior work across PRDs, git history, session names, and work directories with a single command.

Design and integrate web interfaces using Claude Design with optional downstream handoff to frontend-design plugin.

Systematic design review against Nielsen's 10 heuristics and universal design laws. Evaluate UI for usability.

Comprehensive pre-delivery UX audit against all UI/UX standards. Production readiness checklist.

Parallelize work via six patterns: built-in agents, worktree-isolated agents, background agents, custom agents, agent teams, parallel dispatch.

Per-session MCP profile manager with Herdr auto-detect, tmux sync, and Pulse/Conductor integration.

Manages the PAI Knowledge Archive. Typed graph across People, Companies, Ideas, Research with 8 relationship types.

Life OS and project analysis. Manage goals, missions, projects, dependencies, books, movies.

Complete visual content system. Illustrations, diagrams, visualizations, Mermaid flowcharts, header images, technical diagrams.

Programmatic video creation with React. Motion graphics, video rendering, YouTube videos, animated content.

Audio/video editing pipeline. Whisper transcription, Claude classification, ffmpeg editing, optional Cleanvoice polish.

Project-scoped state management across sessions. Persistent tracking via .project/ directory that survives /clear.

Phase-based project roadmapping with dependencies, milestones, and wave execution planning.

Automated pull request creation with branch naming, commit message generation, and PR description templates.

Layered fiction writing using Will Storr's storytelling science and rhetorical figures. Novel, short story, character arc development.

Generate executive summaries from security assessments with automated translation, risk scoring, and quality gates.

Persistent world-model stress-testing ideas across 11 time horizons (6 months–50 years). Geopolitics, tech, economics, society, environment.

Structured incident investigation (5 Whys, Fishbone, Postmortem, Fault Tree, Kepner-Tregoe). Blameless culture, actionable remediations.

Structural analysis of complex systems (Iceberg, Causal Loops, Archetypes, Leverage Points). Meadows, Senge, Forrester methodologies.