Explore skills

Debugs errors and traces failures in AI agents and their tools. Identifies context errors, infinite loops, malformed tool calls, response parsing issues and subagent conflicts.

Runs an autonomous, self-paced "dev loop" that processes a queue of issues/tasks on a git repo. It branches, implements, tests, creates PRs, reviews, merges, verifies deployment, and updates issues, continuing unattended until the queue is exhausted.

Scans projects (JS/TS, Python, etc.) for supply-chain and infostealer patterns. It detects obfuscated JavaScript, malicious loaders, ESM escapes, browser/SSH/cloud credential theft, and reverse shells.

Audits and cleans OpenClaw memory files, removing stale facts, contradicted entries, and outdated context from council_memory.json, MEMORY.md, and daily memory files. Use when memory has accumulated junk, agents are citing old numbers, or before a major project phase change. Run by Samael on Sunday cron.

Versioned, structured memory for AI agents that tracks every fact and logs every change, ensuring your agent always gets the current answer instead of outdated context. Requires MEMSTATE_API_KEY.

TiDB X is object-storage-native distributed SQL for AI agent workloads. Use it when building agent memory, context storage, multi-agent coordination, Web3 indexing, or any durable queryable state for AI systems.

Manage personal subscriptions via Wallos (self-hosted tracker). Use when the user asks to view, add, edit, or delete subscriptions, check spending, manage categories/payment methods, or anything related to subscription tracking and recurring expenses.

Use to set up, install, audit, update, repair, or sync a project-local AI harness for the current repository. It inspects the repo, conducts an interview for defaults and guardrails, then creates or updates synchronized AGENTS.md, CLAUDE.md, GEMINI.md, PROJECT_HARNESS.md, harness-contract.json, and harness-runtime.json files.

Turn a brief, link, or research topic into a controlled, source-grounded video prompt that runs on HiAPI Seedance 2.0 or HappyHorse 1.0. Use when a user asks to direct, plan, or storyboard a video before generation.

Switch between alternative versions of React components live, via a floating pill toolbar. Use when comparing design alternatives in a deployed prototype, or when the user mentions reversion, variant switcher, version toggle, compare versions, A/B preview toolbar, or design switcher.

Use for executing shell commands or external processes requiring platform-aware procedures, federated identity handoff to a real browser session, agent-browser session retry, privilege elevation fallback paths, or destructive file deletion safety. Triggers when invariants in the `command-execution` rule module need procedural how-to detail.

Uses a stealth Playwright browser to search over 18 search engines (e.g., Google, Bing) or crawl any URL. Invoke when web search results are needed to answer a question or when the user explicitly asks to search the web or fetch a webpage.