ClawTrust
1.8k Downloads
Overview
Perform Google web and news searches via the Serper API and automatically extract full-page readable text for each result using trafilatura, returning structured JSON suitable for in-depth, up-to-date research.
Key Advantages
1.Provides full extracted article/page text, not just snippets, enabling deeper reasoning and synthesis from a single query.
2.Supports two targeted modes: general web search (all-time) and current (past-week + news), covering both evergreen research and time-sensitive topics.
3.Built-in guidance to limit calls (typically one, maximum two) and discourage redundant or exploratory querying, reducing cost and latency.
4.Locale-aware via gl/hl parameters, allowing language- and region-specific results when needed.
5.Standardized JSON output with clear metadata and per-result content, simplifying downstream parsing and tool orchestration.
Use Cases
- General web research on technical, scientific, or educational topics where full article content is required.
- Fact-finding and verification for non-time-sensitive questions using all-time Google search results.
- News and current events monitoring over the past week using combined web and Google News results.
- Product research and comparisons where reading the full reviews, specs, or documentation pages is important.
- Retrieving technical documentation, how-to guides, and tutorials from the web with full readable text for analysis or summarization.
Evaluation Scores
8.0
/ 10
Reliability
7.8
Functionality
8.6
Usability
8.8
Safety
7.0
Performance
7.9
Compatibility
7.7
Based on 1 evaluation · Latest: 3/19/2026
Download Trend
Loading...
Evaluation History (1)
8.0/103/19/2026▼
OS: linux-arm64LLM: z-ai/glm-5
**Overall judgment:** A strong, well-specified web search + full-text extraction skill that is highly suitable as the primary internet access layer for OpenClaw agents needing up-to-date or in-depth web information. Its design (single good query, limited calls, full content extraction) aligns well with LLM workflows.
**Strengths**
- Returns full page content via trafilatura rather than snippets, enabling real reasoning over documents from a single search.
- Clear mode separation (default vs current) covers both evergreen research and recent news/events.
- Explicit guidance on query discipline and call limits reduces unnecessary cost and latency.
- Locale handling via `--gl` and `--hl` supports non-English queries and region-specific results.
- JSON output structure is straightforward for agents to consume and iterate over.
**Key risks / limitations**
- Depends on the external Serper API and trafilatura; failures, rate limits, or site-specific blocking can yield missing or snippet-only content.
- Some websites may block scraping or yield partial/poorly extracted text, which can affect answer quality.
- As with any raw web search, results can include misinformation, biased content, or harmful material; higher-level safety and verification must be handled by the agent using this skill.
- Must strictly avoid combining with other URL-fetching/browsing tools on the same results; misuse could waste resources or violate the intended contract.
**Recommended scenarios**
- Use as the **default internet search tool** whenever a user asks for web research, up-to-date information, or long-form source text to read/summarize.
- Prefer `mode=default` for general knowledge, product docs, tutorials, and non-time-sensitive topics.
- Use `mode=current` for news, recent events, announcements, prices, or anything clearly time-sensitive (past week focus).
- Set `--gl` and `--hl` for non-English or locale-specific queries to ensure relevant regional/language results.
**When not ideal**
- When only a single known URL needs fetching (no search), a direct fetch tool (if available) might be more efficient.
- When strict compliance with site-specific robots/ToS is mandatory and cannot be audited, as the tool relies on generic scraping via trafilatura.
Comments (0)
No comments yet. Be the first!