AI & LLM Integration for Claude Code

EU AI Act (Regulation (EU) 2024/1689) Article-cited compliance operator. Three decisions: AI system risk tier (Article 5 / 6+ Annex III / 50 / minimal), conformity assessment routing (Article 43 Module A vs H + Annex IV docs), per-role obligation matrix (provider/deployer/importer/distributor + GPAI). NOT executive AI strategy (see cs-caio-advisor). NOT a legal substitute (engage counsel for novel cases).

ISO/IEC 42001:2023 AI Management System (AIMS) implementation + internal audit operator. Three decisions: AIMS gaps against Clauses 4-10, AI risk register per Annex A + ISO 23894, Clause 9.2 internal audit plan. NOT executive AI strategy (see cs-caio-advisor). NOT EU AI Act conformity (see cs-ai-act-compliance).

Marc Andreessen-mode operator. Runs on a fixed anti-sycophancy operating prompt — leads with the strongest counterargument, never validates premises. Agent-native orchestrator for Claude Code, Codex, Gemini CLI.

Eval-demanding Chief AI Officer advisor for model build-vs-buy decisions, AI risk classification under EU AI Act + US state laws, AI cost economics. Agent-native orchestrator for Claude Code, Codex, Gemini CLI.

Eval-demanding Chief AI Officer advisor for model build-vs-buy decisions, AI risk classification under EU AI Act + US state laws, AI cost economics (API vs self-hosted), and AI team org evolution. Strategic only — does not duplicate engineering AI/ML skills.

Caveman-mode operator. Persistent ultra-compressed communication mode. Drops articles, filler, pleasantries, and hedging while preserving all technical substance. Auto-clarity exception for security warnings, irreversible actions, multi-step sequences, and clarification requests. Activated by user phrases ("caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief") or /cs:caveman command.

Caveman-mode operator. Persistent ultra-compressed communication mode. Drops articles, filler, pleasantries, and hedging while preserving all. Agent-native orchestrator for Claude Code, Codex, Gemini CLI.

Use proactively after any user message in a Claude.ai or Claude Code session where the user is learning to prompt better or has explicitly activated. Agent-native orchestrator for Claude Code, Codex, Gemini CLI.

Use proactively after any user message in a Claude.ai or Claude Code session where the user is learning to prompt better or has explicitly activated coaching. Default persona for the claude-coach skill. Watches each turn for missed power-user opportunities and surfaces at most one ⚡ tip when a tip would genuinely 10x the next attempt.

NotebookLM browser-automation persona. Walks 2-4 forcing intake questions (Q1 action: read / add source / Studio output / create new; Q2-Q4 branch per action). Refuses to start without action commitment. Refuses to handle login flows automatically. Refuses to wait synchronously for Studio generations (fire-and-notify). Always opens Studio customization menu and writes detailed custom prompt (default prompts produce mediocre output). Screenshot-first discipline. find()-before-click. Fails fast wh

NotebookLM browser-automation persona. Walks 2-4 forcing intake questions (Q1 action: read / add source / Studio output / create new; Q2-Q4 branch. Agent-native orchestrator for Claude Code, Codex, Gemini CLI.

Multi-source recency research persona. Walks 2–4 forcing intake questions one at a time (topic specificity, angle, time window, platform scope), runs. Agent-native orchestrator for Claude Code, Codex, Gemini CLI.

Hybrid research router + fallback persona. Walks 2-4 minimal intake questions (Q1 question + Q2 output preference; Q3 disambiguation only when. Agent-native orchestrator for Claude Code, Codex, Gemini CLI.

Hybrid research router + fallback persona. Walks 2-4 minimal intake questions (Q1 question + Q2 output preference; Q3 disambiguation only when classification is ambiguous; Q4 only if fallback). Deterministically classifies research questions by keyword signals and routes to one of 6 specialists (pulse / grants / litreview / syllabus / patent / dossier) at ≥2-signal confidence. Falls back to own plan-decompose-search-synthesize workflow when no specialist matches. NEVER delegates silently — alway

Dispatched sub-agent that ingests a new source into an LLM Wiki vault. Reads the source, proposes TL;DR and key claims, identifies which. Agent-native orchestrator for Claude Code, Codex, Gemini CLI.

Dispatched sub-agent that answers queries against an LLM Wiki vault. Reads index.md first, drills into 3-10 relevant pages across categories. Agent-native orchestrator for Claude Code, Codex, Gemini CLI.

93 agent-native orchestrators for Claude Code, Codex CLI, and Gemini CLI — multi-skill AI agents across engineering, product, marketing, and more.

Persona-Based Agents — agent-native AI orchestrator for Personas. Works with Claude Code, Codex CLI, Gemini CLI, and OpenClaw.

One paragraph describing what this agent does, who it's for, and when to activate it.. Agent-native orchestrator for Claude Code, Codex, Gemini CLI.

Dispatched sub-agent that ingests a new source into an LLM Wiki vault. Reads the source, proposes TL;DR and key claims, identifies which. Agent-native orchestrator for Claude Code, Codex, Gemini CLI.

Dispatched sub-agent that ingests a new source into an LLM Wiki vault. Reads the source, proposes TL;DR and key claims, identifies which entity/concept/synthesis pages will be touched, flags contradictions with existing pages, and — after user confirmation — writes the source summary, updates cross-references across 5-15 pages, regenerates the index, and appends a standardized log entry. Spawn when the user says "ingest this", "add this paper/article/book to the wiki", or drops a file into raw/.

Dispatched sub-agent that answers queries against an LLM Wiki vault. Reads index.md first, drills into 3-10 relevant pages across categories, synthesizes an answer with inline [[wikilink]] citations, and offers to file the answer back into the wiki as a new comparison or synthesis page. Spawn when the user asks a substantive question the wiki might answer, says "what does the wiki say about X", "compare A and B across my sources", or wants to explore a topic.

Dispatched sub-agent that answers queries against an LLM Wiki vault. Reads index.md first, drills into 3-10 relevant pages across categories. Agent-native orchestrator for Claude Code, Codex, Gemini CLI.

Agents are reusable AI components with defined capabilities, tools, and instructions.

ClaudeResearcher Agent Context

每次启动一位领域专家时,把下面模板里的占位符替换掉,作为 prompt 喂给 Agent 工具。

所有专家稿落盘后再启动红队,让它找到最强主张并尝试反驳。

Deduplication judge for the c-review pipeline. Merges duplicate findings deterministically by exact location and bug class, then runs LLM passes over same-function candidates, including the same bug filed under different bug classes. Spawned by the c-review skill orchestrator only.

Deduplication judge for the rust-review pipeline. Merges duplicate findings deterministically by exact location and bug class, then runs LLM passes over same-function candidates, including the same bug filed under different bug classes. Spawned by the rust-review skill orchestrator only.

Use when working on AI system design, model implementation, and production deployment, including multiple AI frameworks and tools, with emphasis on building scalable, efficient, and ethical AI solutions from research to production.

Use when working on statistical analysis, machine learning, and business insights, including exploratory data analysis, predictive modeling, and data storytelling, with emphasis on delivering actionable insights that drive business value.

Use when working on large language model architecture, deployment, and optimization, including LLM system design, fine-tuning strategies, and production serving, with emphasis on building scalable, efficient, and safe LLM applications.

Use when working on production model deployment, serving infrastructure, and scalable ML systems, including model optimization, real-time inference, and edge deployment, with emphasis on reliability and performance at scale.

Use when working on Model Context Protocol server and client development, including protocol specification, SDK implementation, and building production-ready integrations between AI systems and external tools/data sources.

Use when working on machine learning model lifecycle, production deployment, and ML system optimization, including both traditional ML and deep learning, with emphasis on building scalable, reliable ML systems from training to serving.

Use when working on ML infrastructure, platform engineering, and operational excellence for machine learning systems, including CI/CD for ML, model versioning, and scalable ML platforms, with emphasis on reliability and automation.

Use when working on natural language processing, understanding, and generation, including transformer models, text processing pipelines, and production NLP systems, with emphasis on multilingual support and real-time performance.

Use when designing, optimizing, and managing prompts for large language models, including prompt architecture, evaluation frameworks, and production prompt systems, with emphasis on reliability, efficiency, and measurable outcomes.