Total 50,604 skills, AI & Machine Learning has 8484 skills
Showing 12 of 8484 skills
Use when user wants to execute long-running tasks that require multiple sessions to complete. This skill manages task decomposition, progress tracking, and autonomous execution using Codex non-interactive mode with auto-continuation. Trigger phrases include autonomous, long-running task, multi-session, 自主执行, 长时任务, autonomous skill.
Generate images with Alibaba Cloud Model Studio Z-Image Turbo (z-image-turbo) via DashScope multimodal-generation API. Use when creating text-to-image outputs, controlling size/seed/prompt_extend, or documenting request/response mapping for Z-Image.
Refresh the Model Studio models crawl and regenerate derived summaries and `skills/ai/**` skills. Use when the models list or generated skills must be updated.
Use when the user asks to "find X in codebase", "where is Y defined?", "explore this dir", "list files in src/", "trace definitions", "find usages" — local-only. Local codebase exploration via Octocode Local + LSP. No GitHub; for external repos use octocode-research.
Model Context Protocol development expert. Use when creating MCP servers, clients, or tools that enable AI agents to interact with external systems, APIs, and development environments.
Use this skill to query your Google NotebookLM notebooks directly from Claude Code for source-grounded, citation-backed answers from Gemini. Browser automation, library management, persistent auth. Drastically reduced hallucinations through document-only responses.
Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward model training. Use when need RLHF, align model with preferences, or train from human feedback. Works with HuggingFace Transformers.
Coordinate complex work using a phase-gated, multi-agent engineering loop (audit → design → implement → review → validate → deliver). Use when you need to split a task into subsystems, run dual independent audits, reconcile findings into a confirmed issue list, delegate fixes in clusters, enforce dual-review PASS gates, and drive an end-to-end delivery. Prefer discovering and invoking other specialized skills when they can execute part of the work faster or more reliably.
Use after completing work sessions to analyze agent behavior patterns, prepare session handoffs for continuity, document completed work, identify blockers, or preserve context for the next session.
Search group Yuque knowledge bases with natural language queries and provide summarized answers with key points and source links. For group use — searches within team/group repositories. Requires group Token.
Use when brainstorming, evaluating architecture choices, or comparing trade-offs where independent perspectives from different model families (Claude/Codex/Gemini) would surface blind spots
Scaffold a new integration-type "seed skill" — a SKILL.md with reference assets and prompts.jsonl entries that teach an agent to generate working, testable integration code from scratch. - Use when: developer wants to create a new skill for an SDK or framework integration that agents should be able to implement and test - Ask first: Is this an integration skill (code generation + tests) or a tool skill (CLI wrapper with no tests)? Only proceed with this skill for integration-type skills.