Total 58,125 skills, AI & Machine Learning has 9662 skills
Showing 12 of 9662 skills
Local-first, security-first control center for OpenClaw agents — visibility dashboard with readonly defaults, token attribution, collaboration tracing, and safe write operations.
Google Model Armor: Create a new Model Armor template.
Auto-generates an LLM usage monitoring page in a PM admin dashboard. Tokuin CLI-based token/cost/latency tracking + user ranking system + inactive user tracking + data-driven PM insights + Cmd+K global search + per-user drilldown navigation. Supports OpenAI/Anthropic/Gemini/OpenRouter.
Creates and configures Claude Code hooks for lifecycle automation. Covers all 17 hook events, 4 hook types (command, prompt, agent, http), matchers, input/output formats, and exit codes. Follows official Anthropic best practices. USE WHEN: user mentions "hook", "hooks", "auto-format", "pre tool use", "post tool use", "session start", "notification hook", "block command", "validate tool", "lifecycle event", "PostToolUse", "PreToolUse" DO NOT USE FOR: creating skills - use `skill-authoring`; creating agents - use `agent-authoring`; webhook endpoints - different concept
Manage OpenCode's permission rules in opencode.jsonc — add, remove, or list auto-approval rules for Bash commands and tool invocations so the agent stops asking for confirmation on every single command. Use whenever the user wants to auto-approve, deny, or require confirmation for a shell command, even if they don't mention "permission" or "opencode.jsonc" directly. Triggers on "允许 kubectl get *", "拒绝 rm -rf", "auto-approve npm run build", "总是执行 git status", "add permission rule", "list my permissions", "查看权限", "添加权限", "移除权限", "把 X 加到允许列表", "skip confirmation for", and similar — even if the user doesn't explicitly mention OpenCode's config.
Connect OpenClaw AI agents to personal WeChat accounts for messaging, group chats, and automation
Orchestrate Devin CLI subagents as background workers using tmux windows. Use when the user asks to spawn, coordinate, fan-out, or delegate work to multiple parallel agents, run background Devin sessions, or orchestrate long-running autonomous tasks from inside an existing Devin session.
Local Skill Cleaner. Scans skills in Claude Code, Codex, Grok, general Agents and specified directories to identify content that violates user authorization, such as advertising diversion, hidden commercial intent, task hijacking, suspicious external calls, sensitive data reading, etc.; by default, it only generates a report and quarantines problematic skills after user confirmation. Trigger methods: /dbs-skill-cleaner, /clean skills, /check skills, "scan local skills", "detect skill ads", "remove problematic skills", "review my skills" Local skill cleaner. Scans installed or specified skills for advertising, covert commercial intent, task hijacking, suspicious external calls, and sensitive-data access. Reports first and quarantines only after explicit confirmation. Trigger: /dbs-skill-cleaner, /clean skills, /check skills, "scan my local skills", "detect skill ads", "clean problematic skills"
Use when the user asks to "create a metric", "write a metric", "design a metric", "build a metric for", "evaluate agent performance", "measure call quality", "track a KPI", "add a workflow metric", "improve my metric", "fix a metric", "debug metric results", "set up quality scoring", or "what metrics do I need". Also relevant when discussing LLM judge prompts, custom code metrics, evaluation triggers, VALID_SKIP patterns, section extraction, or metric best practices for Cekura voice AI agents. Covers both creating new metrics and reviewing, iterating on, or troubleshooting existing ones.
Use this when the user requests you to execute multiple tasks in parallel, start multiple workers/agents simultaneously, launch multiple independent sessions using tmux, prevent PM from directly bypassing implementation protocols, or when you act as a PM to decompose and assign tasks to multiple independent workers. Trigger terms include "parallel execution", "start multiple", "simultaneous execution", "assign workers", "multi-agent parallelism", "start workers", "tmux launch", "independent session", "anti-escape", "task assignment", "do together". Do not use for single short tasks, cross-platform task status management, or Git branch/commit/PR/merge security rules.
High-performance reinforcement learning framework optimized for speed and scale. Use when you need fast parallel training, vectorized environments, multi-agent systems, or integration with game environments (Atari, Procgen, NetHack). Achieves 2-10x speedups over standard implementations. For quick prototyping or standard algorithm implementations with extensive documentation, use stable-baselines3 instead.
Google quantum computing framework. Use when targeting Google Quantum AI hardware, designing noise-aware circuits, or running quantum characterization experiments. Best for Google hardware, noise modeling, and low-level circuit design. For IBM hardware use qiskit; for quantum ML with autodiff use pennylane; for physics simulations use qutip.