NLP

Unified documentation and context preservation: spec folder workflow (levels 1-3), template enforcement, validation, Spec Kit Memory with vector search, six-tier importance system, constitutional rules, checkpoint save/restore. Mandatory for all file modifications.

Test cross-model context handoff integrity.

Master coordinator for Unite-Hub workflows. Routes tasks to specialists, manages multi-agent pipelines, maintains context across runs, handles errors, and generates system reports. The brain of the automation system.

Understand authentication and authorization defects in AI-generated code including insecure password storage, broken session management, and access control bypasses. Use this skill when you need to learn about auth vulnerabilities in AI code, understand why AI suggests MD5/plaintext passwords, recognize broken session patterns, or identify access control gaps. Triggers include "auth vulnerabilities AI", "password storage AI", "session management", "broken access control", "authentication defects", "MD5 passwords", "session hijacking", "authorization bypass".

Design and implement state management solutions using Context API, XState, Zustand, Jotai, and custom hooks with testing patterns and performance optimization

Work with Resource datasets (mutable state tracking) using OPAL temporal joins. Use when you need to enrich Events/Intervals with contextual state information, track resource state changes over time, or navigate between datasets using temporal relationships. Covers temporal join mechanics (lookup, join, follow), automatic field matching, and when to use Resources vs Reference Tables.

Expert at integrating Model Context Protocol (MCP) servers into Claude Code plugins. Auto-invokes when the user wants to add external tool integrations, configure MCP servers, set up stdio/SSE/HTTP/WebSocket connections, or needs help with MCP authentication and security. Also auto-invokes proactively when Claude is about to write MCP configuration files (.mcp.json) or add mcpServers to plugin manifests.

Filter and search event datasets (logs) using OPAL. Use when you need to find specific log events by text search, regex patterns, or field values. Covers contains(), tilda operator ~, field comparisons, boolean logic, and limit for sampling results. Does NOT cover aggregation (see aggregating-event-datasets skill).

Executes Exa AI search queries via unified Python CLI. Use when searching the web for current information, finding code examples, researching APIs, SDKs, or retrieving programming context for any library or framework.

Integrate Gemini API with correct current SDK (@google/genai v1.27+, NOT deprecated @google/generative-ai).Supports text generation, multimodal (images/video/audio/PDFs), function calling, and thinking mode. 1M input tokens.Use when: integrating Gemini API, implementing multimodal AI, using thinking mode for reasoning, function callingwith parallel execution, streaming responses, deploying to Cloudflare Workers, building chat, or troubleshootingSDK deprecation, context window, model not found, function calling, or multimodal format errors.Keywords: gemini api, @google/genai, gemini-2.5-pro, gemini-2.5-flash, gemini-2.5-flash-lite,gemini-3-pro-preview, multimodal gemini, thinking mode, google ai, genai sdk, function calling gemini,streaming gemini, gemini vision, gemini video, gemini audio, gemini pdf, system instructions,multi-turn chat, DEPRECATED @google/generative-ai, gemini context window, gemini models 2025,gemini 1m tokens, gemini tool use, parallel function calling, compositional function ca

Build a JIT Modelfile that hardcodes current context into SYSTEM/TEMPLATE/PARAMETER and creates a temporary Ollama model. Use when you need a hyper-focused ephemeral model.

Use when design is complete and you need detailed implementation tasks for engineers with zero codebase context - creates comprehensive implementation plans with exact file paths, complete code examples, and verification steps assuming engineer has minimal domain knowledge | 設計が完了し、コードベースの知識がゼロのエンジニア向けに詳細な実装タスクが必要な場合に使用 - 正確なファイルパス、完全なコード例、検証ステップを含む包括的な実装計画を作成。エンジニアの領域知識が最小限であることを前提

Access Context Portal (ConPort) as persistent project memory via MCPorter. Load/update project context, log decisions and progress, track patterns. Use PROACTIVELY at session start to load context and throughout work to log decisions. Combats AI amnesia across sessions.

Lightweight Linear skill for fetching issue details. Returns only essential data (title, description, state, assignee) to optimize context usage compared to full Linear MCP.

Transform instructions that tell what not to do into positive guidance toward what to do, since negation leaves an unbounded space of prohibited actions while positive instruction provides specific direction. Use when: (1) reviewing any instructional text containing "don't," "avoid," "never," or "do not", (2) constraints in a skill feel arbitrary without explanation, (3) instructions describe boundaries without providing direction.

Use when analyzing requirements or project specifications - guides shannon analyze command, explains 8D complexity output, caching behavior, context-aware mode with --project flag

Persist data with Core Data - models, contexts, fetch requests, migrations, SwiftData

Comprehensive AI writing detection patterns and methodology. Provides vocabulary lists, structural patterns, model-specific fingerprints, and false positive prevention guidance. Use when analyzing text for AI authorship or understanding detection patterns.

Create searchable knowledge base from existing documentation and code for fast context lookup in brownfield projects. Indexes .md files with sections/keywords, maps code exports to documentation, generates search.json, quick-ref.md, and glossary.md in .claude/index/. Use when onboarding to brownfield projects or when documentation needs better discoverability.

Sync Linear cycle data with CURRENT-FOCUS.md, analyze chat context for gaps, prioritize work based on dependencies and test failures, and provide actionable recommendations. This skill should be used when the user requests "show current focus", "update current focus", or discusses project status and upcoming work.