🎁 Get the FREE AI Skills Starter Guide β€” Subscribe β†’
BytesAgainBytesAgain
BytesAgain β€Ί Skills β€Ί Local Web Search
πŸ¦€ ClawHub

Local Web Search

by @psanger

Real-time web search for any OpenClaw commander model. Default path is free/private local SearXNG + Scrapling/browser-worker search with no API keys; optiona...

Versionv4.2.0
#requires-sensitive-credentials#clawhub
πŸ”’ Constraints

  • Always use --intent to match the query type for best results. --intent is part of this skill's own workflow, not a universal OpenClaw flag. Agents that read/follow this skill should choose it automatically from task type, but agents that do not load the skill will not automatically inherit these conventions.
  • When local SearXNG is unavailable, scripts can optionally use LOCAL_SEARCH_FALLBACK_URL if you set it explicitly.
  • If the fallback also fails, tell the user to start local SearXNG:

    • cd "$(cat ~/.openclaw/workspace/skills/local-web-search/.project_root)" && ./start_local_search.sh

  • Do NOT invent search results if all sources fail.
  • search_local_web.py and browse_page.py are complementary: search first, browse second.
  • Prefer [cross-validated] results (appeared in multiple engines) for factual claims.
  • For sites behind Cloudflare or requiring JS, use browse_page.py --mode stealth.
  • If BROWSER_WORKER_URL is set, browse_page.py will delegate auto, stealth, and dynamic modes to that worker. Keep this for environments where the worker is actually reachable, such as inside the same Docker network or through an intentional tunnel/proxy.
  • For specific factual claims (dates, numbers, names, events), use verify_claim.py to get a multi-source confidence score before asserting.
  • Never assert a claim with UNCERTAIN, LIKELY_FALSE, or UNVERIFIABLE verdict β€” tell the user the evidence is insufficient instead.
  • This skill works identically regardless of which LLM model is acting as the OpenClaw commander. No model-specific behavior is assumed.

    • View on ClawHub
    TERMINAL
    clawhub install local-web-search

    πŸ§ͺ Use this skill with your agent

    Most visitors already have an agent. Pick your environment, install or copy the workflow, then run the smoke-test prompt above.

    Manus
    Task-oriented agent. Great for testing AI skills end-to-end.
    Try Manus β†’
    OpenClaw
    Local-first agent. Install skills via ClawHub CLI.
    Set up OpenClaw β†’
    Claude Code
    Anthropic's coding agent. Paste the prompt or SKILL.md into your session.
    Claude Code docs β†’
    Cursor
    AI-powered IDE. Use the smoke-test prompt in Cursor Agent.
    Open Cursor β†’
    Continue.dev
    Open-source AI code assistant. Add SKILL.md as a custom tool.
    Continue docs β†’
    Windsurf
    Agentic IDE by Codeium. Paste the prompt into Cascade.
    Try Windsurf β†’
    Cline
    VS Code extension for autonomous coding with MCP tools.
    Cline on GitHub β†’
    Copilot Workspace
    GitHub's AI dev environment. Suitable for code-generation skills.
    Copilot Workspace β†’

    πŸ” Can't find the right skill?

    Search 60,000+ AI agent skills β€” free, no login needed.

    Search Skills β†’