Synthesize claims across multiple sources to identify consensus, disagreements, and emerging narratives on AI research topics. Use when you have claims from both lab researchers and critics on the same topic and need to understand where they agree, disagree, and what the overall hype level is.
Specialized agent for deep argument analysis. Maps claims, identifies logical gaps, generates counterarguments, suggests evidence needs. Use when user asks to "strengthen arguments", "check logic", "improve reasoning", or explicitly invokes the argument strengthener agent.
Use when user wants to ask Codex a question, get a second opinion, or consult with gpt-5.2. Triggers on phrases like "ask codex", "codex opinion", "what does codex think".
PKE (Promethean Knowledge Engine) query patterns and memory framework integration. Use for: (1) Structured knowledge retrieval, (2) Context package assembly, (3) Course spec lookups, (4) Terminology and formatting rules, (5) Semantic anchoring. Triggers: PKE, knowledge engine, memory, context, retrieval, query, lookup, course spec, terminology.
Get second opinions from external AI agents (OpenAI, Gemini, GitHub). Use when user says "second opinion", "ask other AIs", "what would GPT say", "check with others", or wants to validate an approach.
Orchestrate comprehensive TODAS research for novel/emerging domains (1-7 subagents adaptive). Specializes in unprecedented topics, post-training data, and emerging technologies. Uses adaptive depth-based methodology: straightforward queries (1 agent), standard queries (2-3 agents), complex queries (5-7 agents). Handles depth-first (multiple perspectives), breadth-first (distinct sub-topics), and straightforward investigations. Triggers include "novel", "emerging", "2025", "2026", "unprecedented", "new technology", research on topics that didn't exist during training cutoff.
Browser Controller拡張機能を使ってChatGPT 5.2 Thinkingで3並列以上のウェブ検索・ブレスト・情報収集を実行する。難問にはHeavy thinking（heavy/extended）を使用。指定時のみChatGPT Proで深い考察・ファクトチェック。マルチターン継続（chat）、再取得（recover）、ファイル添付、タブ管理にも対応。「ウェブ検索」「横断検索」「並列検索」「ブレスト」「情報収集」「リサーチ」「再取得」「追加質問」を依頼されたときに使用する。
Search Claude conversation history from JSONL files. Use when looking for previous discussions, past decisions, code solutions, or context from earlier conversations.
Deep web research, structured answers, and content retrieval using Exa. Use for heavy research, multi-step synthesis, or rich structured outputs; optionally pair with brave-search for fast scoping but not required.
Reconstruct, visualize, and analyze argument structure. Use for: argument reconstruction, premise identification, inference evaluation, finding hidden assumptions, visualizing debates, Toulmin model analysis. Triggers: 'argument structure', 'premises', 'conclusion', 'inference', 'reconstruct', 'map the argument', 'Toulmin', 'argument diagram', 'validity', 'soundness', 'implicit premise', 'hidden assumption', 'logical structure'.
Document Q&A with RAG using Supabase pgvector store.
A tiered approach to answering research questions with clear stop rules, evidence capture, and escalation to Tavily/Playwright only when needed.
Evaluation framework patterns for RAG and LLMs, including faithfulness metrics, synthetic dataset generation, and LLM-as-a-judge patterns. Triggers: ragas, deepeval, llm-eval, faithfulness, hallucination-check, synthetic-data.
Use when needing Claude bot review recommendations from PR - automatically waits for workflows (including Claude review) to complete, then extracts recommendations and returns concise summary without loading full review into context
Analyze English lesson materials and add comprehensive notes including topic summaries and advanced vocabulary (英検2級以上). Use when the user asks to review lesson materials, analyze articles from the lesson, or add vocabulary notes after an English conversation class.
Comprehensive research combining web search, quantum analysis, and persistent memory. Use when researching topics that need current information plus deep cognitive analysis.
Delegates research to Google Gemini. Use when user says "이거 좀 찾아봐", "Gemini한테 물어봐", "문서 검색해줘", "best practice 알려줘", "리서치해줘", "최신 정보 찾아줘", or needs web search, documentation lookup.
Perplexity Sonar API development with search-augmented generation, real-time web search, citations, and OpenAI-compatible Chat Completions. Use for AI-powered applications requiring up-to-date information, research assistants, and grounded responses with sources.
Prepare synthesis guidance for the-art. NOT an outline. Provides thematic guidance, narrative intent, and evidence library that enables synthesis instead of expansion.
Multi-model adversarial debate where Claude defends positions against external CLI models (Gemini, Codex, Qwen). Use when facing complex decisions, wanting to expose blind spots, or needing documented reasoning for tradeoffs. Auto-triggers: "debate this", "challenge my thinking", "what am I missing", "devil's advocate", "stress test this idea", "poke holes", "second opinion", "validate this decision", "am I wrong about"
Search previous Claude Code conversations for facts, patterns, decisions, and context using semantic or text search
Verifies relationships between a source text (e.g. Episode, Task) and existing Knowledge Graph lessons. Runs KNN/Hybrid search to find candidates, then uses LLM (scillm) to verify "verifies", "contradicts", or "related" stances with strict rationales.
Use for deep multi-step investigation and analysis with both Codex and Gemini. Triggers on "council deep analysis", "investigate with council", "council thorough investigation", "have the council think deeply".
