agentleFS
NVIDIAKnown publisherNVIDIA/nemoclaw-community/examples/recipes/nvidia/developer-community-chief-of-staff/agents/hermes/skills/public-web-search/SKILL.md

public-web-search

Skill142 starsChanged 3 months agoApache-2.0View on GitHub

Search the current public web through the recipe's optional, policy-scoped Tavily integration and return cited titles, URLs, snippets, and search metadata. Use for current public information that is not available from the configured Slack, Outlook, GitHub, or source-ETL skills. Do not use for opening a result page, extracting a URL, browser automation, or arbitrary HTTP requests.

  • Reads credentials

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 2 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use Hermes's native web_search tool for current public-web discovery.

    line 9
  2. Very specific:

    Use web_search only. Do not use web_extract, a browser tool, web_fetch, curl, wget, Python HTTP code, or a user-supplied URL as a fetch …

    line 14
  3. Very specific:

    Do not inspect, print, transform, or troubleshoot TAVILY_API_KEY.

    line 18
  4. Distinctive:

    Do not claim that a snippet proves facts beyond the returned title, URL, snippet, score, or other search metadata.

    line 20
  5. Very specific:

    Call web_search. Do not follow result URLs with another tool.

    line 27
  6. Distinctive:

    Prefer primary and authoritative sources in the returned results.

    line 28
  7. Very specific:

    If the tool is absent or reports that Tavily is unavailable, stop.

    line 31

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Public Web Search

Use Hermes's native web_search tool for current public-web discovery. Treat each result as search-provider evidence, not as verified page text.

Enforced boundary

  • Use web_search only. Do not use web_extract, a browser tool, web_fetch, curl, wget, Python HTTP code, or a user-supplied URL as a fetch target.
  • The OpenShell provider and sandbox policy permit only POST /search on api.tavily.com from Hermes's pinned Python runtime.
  • Do not inspect, print, transform, or troubleshoot TAVILY_API_KEY. The sandbox receives an OpenShell placeholder, not the raw API key.
  • Do not claim that a snippet proves facts beyond the returned title, URL, snippet, score, or other search metadata.

Search procedure

  1. Convert the request into one focused search query. Add a date, product name, or official-domain qualifier when it improves precision.
  2. Call web_search. Do not follow result URLs with another tool.
  3. Prefer primary and authoritative sources in the returned results.
  4. Cite the returned URL next to every claim that depends on a result. Label a statement as a search-result snippet when the page itself was not read.
  5. If the tool is absent or reports that Tavily is unavailable, stop. Say: Public web search is disabled for this sandbox. Configure TAVILY_API_KEY on the host, then recreate the sandbox. Do not try another network path.

Output

Return a compact result list with the title, source URL, snippet, and useful provider metadata. Separate direct result metadata from your inference.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.