LLM & Agents

Building Expo DevTools Plugins with CLI Interfaces for interacting with running Expo apps using agents.

Configure LLM providers, use fallback models, handle streaming, and manage model settings in PydanticAI. Use when selecting models, implementing resilience, or optimizing API calls.

Manage GitHub repositories for Claude Code plugins including issues, pull requests, releases, CI/CD workflows, and GitHub Actions. Use when working with GitHub repository management, creating releases, setting up CI/CD, managing issues or pull requests.

Vercel AI Elements for workflow UI components. Use when building chat interfaces, displaying tool execution, showing reasoning/thinking, or creating job queues. Triggers on ai-elements, Queue, Confirmation, Tool, Reasoning, Shimmer, Loader, Message, Conversation, PromptInput.

Expert knowledge for building AI agents with Argentic - a Python microframework for async MQTT-based agents with multi-LLM support, custom tools, and multi-agent orchestration

This skill should be used when the user asks to "set up a team", "create a swarm", "spawn teammates", "assign tasks", "coordinate agents", "work in parallel", "divide work among agents", "orchestrate multiple agents", or describes a complex task that would benefit from multiple Claude Code instances working together. Provides comprehensive guidance for team leads on creating teams, spawning teammates, assigning work, and monitoring progress across tmux/kitty terminal multiplexers.

LLM function calling and tool use patterns. Use when enabling LLMs to call external tools, defining tool schemas, implementing tool execution loops, or getting structured output from LLMs.

Reviews Deep Agents code for bugs, anti-patterns, and improvements. Use when reviewing code that uses create_deep_agent, backends, subagents, middleware, or human-in-the-loop patterns. Catches common configuration and usage mistakes.

Execute feature plans from ~/.ai/plans using progressive disclosure and state tracking. This skill reads existing feature plans, tracks implementation progress in implementation-state.md, uses the Plan agent to explore plans and codebases, and systematically implements features phase by phase. Pairs with the feature-planning skill.

Redis semantic caching for LLM applications. Use when implementing vector similarity caching, optimizing LLM costs through cached responses, or building multi-level cache hierarchies.

Extracts web content and performs reliable searches when standard tools fail due to access restrictions, rate limiting, or validation errors. Use when encountering 403/429/422 errors, blocked documentation sites, or silent search failures. **Direct tool execution often fails for these scenarios - search-plus agent delegation provides the most reliable results.**

Provider-native prompt caching for Claude and OpenAI. Use when optimizing LLM costs with cache breakpoints, caching system prompts, or reducing token costs for repeated prefixes.

This skill should be used before committing code to ensure it meets Breenix quality standards. Use for running clippy checks, fixing compiler warnings, verifying no log side-effects, checking for dead code, and enforcing project coding standards from CLAUDE.md.

Research web content, documentation, and best practices using the gemini-search agent for token-efficient searches with caching. Use when you need to fetch documentation, research best practices, or gather information from the web.

Agentic workflow patterns for autonomous LLM reasoning. Use when building ReAct agents, implementing reasoning loops, or creating LLMs that plan and execute multi-step tasks.

Develops and administers SAP Build Work Zone, advanced edition digital workplace solutions. Use when creating workspaces, workpages, and collaborative sites, developing UI Integration Cards in SAP Business Application Studio, building content packages and workspace templates, integrating with Microsoft 365/Teams/SharePoint/Google Drive, configuring chatbots and webhooks, implementing SCIM API user provisioning, setting up OData business records, managing themes and branding, configuring role-based access and SSO, troubleshooting deployment issues, or working with the Administration Console. Keywords: SAP Build Work Zone advanced edition, digital workplace, UI Integration Cards, content packages, workspace templates, SAP Business Application Studio, SAP Conversational AI, SCIM API, OData, Microsoft Teams integration, SSO, theming, Administration Console

Orchestrate multiple AI instances (clauded/codex CLIs) in tmux panes. Self-discovering coordinator with mandatory verification, synchronous monitoring, and auto-approval. Zero hallucination via imperative commands. (project, gitignored)

Auto-loads when CLAUDE_CODE_IS_TEAM_LEAD environment variable is set. Provides guidance for spawned team-leads on monitoring teammates, handling consults, assigning tasks, and coordinating the team. Use this when you are a spawned team-lead (not the original orchestrator who created the swarm).

Use when creating feature specifications after brainstorming - generates lean spec documents that reference constitutions heavily, link to external docs instead of embedding examples, and focus on WHAT not HOW (implementation plans handled separately)

AI agent development standards including frontmatter structure, naming conventions, tool access patterns, model selection, and Bash-only file operations for .claude/ folders