Mxy Web Search

Web search without an API key using DuckDuckGo Lite via web_fetch. Use as a fallback when web_search fails with missing_brave_api_key error, or whenever you...

Security Scan

The skill's instructions, requirements, and behavior are coherent with its stated purpose (performing DuckDuckGo Lite web searches via web_fetch); it requires no credentials or installs and is instruction-only.

Purpose & Capability

The name/description (DDG Lite web search fallback) matches the SKILL.md instructions: calling web_fetch on https://lite.duckduckgo.com/lite/?q=QUERY and fetching result pages. No unrelated binaries, credentials, or installs are requested.

Instruction Scope

Instructions stay within search/fetch scope: perform a DDG Lite query, parse titles/snippets/URLs, then optionally fetch linked pages. Note: fetching linked pages means the agent will retrieve arbitrary third-party webpages (expected for a search skill) — this can surface sensitive content or cause network requests to external sites.

Install Mechanism

No install spec (instruction-only), so nothing is written to disk or downloaded. This is the lowest-risk install model and matches the stated zero-dependency claim.

Credentials

The skill declares no environment variables, credentials, or config paths — consistent with a simple web-fetch-based search. There are no unexplained secret requests.

Persistence & Privilege

always:false and model invocation not disabled (defaults). The skill does not request elevated persistence or system-wide config changes; autonomous invocation is normal for skills.

Assessment

This skill appears to do exactly what it says: perform DuckDuckGo Lite searches using the agent's web_fetch tool and optionally fetch result pages. No credentials or installs are required. Things to consider before installing: 1) Privacy — queries and fetched pages will be sent over the network to DuckDuckGo and to any linked sites; avoid sending sensitive secrets in queries. 2) Auditing — the package has no homepage/source and the _meta.json ownerId differs from the registry owner id shown in metadata, which reduces provenance; if provenance matters, ask the publisher for source or a repo. 3) Fetch behavior — because the skill may fetch arbitrary result URLs, it can retrieve third-party content; if you want to limit risk, keep it user-invocable only (it already is) and avoid enabling autonomous invocation in high-sensitivity contexts. Overall this is internally coherent and low-risk, but verify your environment's web_fetch privacy/logging policy if network exposure is a concern.

DuckDuckGo Search via web_fetch

Search the web using DuckDuckGo Lite's HTML interface, parsed via web_fetch. No API key or package install required.

How to Search

web_fetch(url="https://lite.duckduckgo.com/lite/?q=QUERY", extractMode="text", maxChars=8000)

• URL-encode the query — use + for spaces
• Use extractMode="text" (not markdown) for clean results
• Increase maxChars for more results

Region Filtering

Append &kl=REGION for regional results:

• au-en — Australia
• us-en — United States
• uk-en — United Kingdom
• de-de — Germany
• fr-fr — France

Full list: https://duckduckgo.com/params

Example — Australian search

web_fetch(url="https://lite.duckduckgo.com/lite/?q=best+coffee+melbourne&kl=au-en", extractMode="text", maxChars=8000)

Reading Results

Results appear as numbered items with title, snippet, and URL. Skip entries marked "Sponsored link" (ads) — organic results follow.

Search-then-Fetch Pattern

1. Search — query DDG Lite for a list of results
2. Pick — identify the most relevant URLs
3. Fetch — use web_fetch on those URLs to read full content

Tips

• First 1-2 results may be ads — skip to organic results
• For exact phrases, wrap in quotes: q=%22exact+phrase%22
• Add specific terms to narrow results (site name, year, location)

Limitations

• No time/date filtering (DDG Lite doesn't support &df= reliably via fetch)
• Text results only — no images or videos
• Results sourced from Bing (may differ from Google)
• Google search does NOT work via web_fetch (captcha blocked)