Skills for your AI
Designs and builds specialized Claude Code agents in Markdown for the claude-code-templates system, covering domain analysis, specification writing, prompt engineering, and template patterns. Use when the user wants to create a new agent, define its expertise
Orchestrates teams of specialized AI agents with roles, tasks, tools, and process types to collaborate on complex tasks. Use when defining agents, sequencing tasks, running a crew, generating YAML configs, assigning tools, or choosing sequential vs hierarchica
Creates, improves, tests, and packages custom skill files for an AI runtime, including trigger optimization and eval suites. Use when the user wants a new skill built from a spec, an existing skill file refined, an eval run on a skill, description triggers opt
Integrates the Opik client, governs prompts and experiments, manages workspaces and projects, and investigates traces, metrics, and quality gates. Use when onboarding a repo to Opik, versioning prompts, organizing telemetry, debugging latency or cost anomalies
Creates reusable Claude Code slash command markdown files in .claude/commands/ or ~/.claude/commands/ from a user's workflow. Use when the user wants to turn a workflow into a slash command, design or write a custom command, or decide between project-level and
Creates, edits, and organizes Claude Code slash commands as markdown files with YAML frontmatter, and explains command features. Use when the user wants a new slash command, wants to modify or namespace existing commands, or asks about $ARGUMENTS, @ file refer
Turns rough copyright ideas into structured AI prompts for research, drafting, and protection workflows. Use when narrowing a copyright topic, choosing an action word, adding context, specifying output format, or refining a weak AI response.
Turns copyright questions into focused, action-oriented AI prompts with context and output specs. Use when narrowing a broad copyright theme, crafting or refining a copyright prompt, adding background such as jurisdiction or doctrine, or specifying tone, langu
Use proactively whenever LLM API costs come up -- or should. Triggers include: 'my AI costs are too high', 'optimize token usage', 'which model should I use', 'LLM spend is out of control', 'implement prompt caching', 'we're about to launch an AI feature', 'bu
Designs and runs adversarial evaluations of deployed LLMs — jailbreak probes, prompt-injection harnesses, and output-safety measurement — producing scored results and signed evidence bundles. Use when planning a red-team engagement, building or running a probe
Generates music, sound effects, and conditioned audio from text descriptions using AudioCraft models. Use when the user asks to create a music clip, sound effect, stereo track, melody- or style-conditioned track, or to continue an existing audio clip.
Analyzes code names for vagueness, misleading behavior, unclear abbreviations, boolean and magic-number issues, and convention violations, then suggests better names and produces a structured report. Use when reviewing naming in a file, snippet, or directory,
Guides owners through customizing a NanoClaw assistant by planning changes to channels, MCP integrations, behavior, commands, and deployment, and providing the exact steps and commands to run. Use when the owner wants to add a channel or MCP integration, chang
Implements the NOWAIT technique for efficient reasoning in R1-style LLMs by suppressing self-reflection tokens to cut chain-of-thought length 27-51% while preserving accuracy. Use when assessing model suitability, configuring the logit processor for HuggingFac
Guides design, composition and optimization of DSPy pipelines — signatures, modules, optimizers, LM providers, RAG and agents. Use when the user asks to build or improve a DSPy program, define a signature, pick a module, configure an optimizer or LM provider,
Constrains LLM output with regex patterns, selection lists, and grammars to guarantee valid JSON, XML, or code, and builds multi-step guided workflows. Use when the user needs format-guaranteed generation, multiple-choice classification, structured JSON/XML ou
Teaches developers to extract validated structured data from LLM responses with the Instructor library, covering setup, Pydantic models, validation and retries, streaming, and extraction examples. Use when the user asks about Instructor setup, response models,
Guides users to set up guaranteed-valid structured generation with the Outlines library for local LLMs, covering Pydantic JSON, choice, regex, integer, float, and nested models. Use when a user wants valid JSON, XML, regex, or code output from a local model, n
Drafts and maintains concise repo-specific instruction files for coding agents, covering initial drafts, updates after agent mistakes, and nested-file advice. Use when creating an agent instruction file for a repository, correcting one after a repo-specific mi
Use when the user asks to optimize prompts, design prompt templates, evaluate LLM outputs with an eval set, measure RAG retrieval quality, validate agent/tool configurations, analyze token usage, or design structured-output contracts. Covers eval-driven prompt
Optimizes prompts, designs RAG, agent, evaluation, production, and security setups for LLM systems, returning annotated rewrites, configuration changes, and design documents. Use when improving a prompt, fixing RAG relevance, designing an agentic workflow, def
Creates, tests, evaluates, and iteratively improves SKILL.md skills for an AI runtime, including trigger-description optimization. Use when the user wants to build a new skill, edit an existing one, write test prompts, review eval results, or improve a skill's
Builds labelled eval sets, sweeps criteria wordings and thresholds, and calibrates confidence gates for typed-decision models. Use when evaluating a typed-decision model, comparing criteria configs or models, choosing a score threshold, testing a confidence ga
Optimizes AI agent performance by pruning redundant context, managing token usage, and enforcing ultra-concise, direct-to-value responses.
