Tavily
Security checks across static analysis, malware telemetry, and agentic risk
Overview
This appears to be a normal Tavily web-search connector; it uses your Tavily API key and sends search, extraction, and research requests to Tavily.
This skill is reasonable for Tavily-based web search and extraction. Before installing, confirm you are comfortable providing a Tavily API key, sending query and URL data to Tavily, and running the bundled Python CLI. Keep searches scoped, avoid confidential data in prompts, monitor credit usage, and treat extracted web page text as untrusted source material rather than instructions.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
VirusTotal findings are pending for this skill version.
Risk analysis
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Anyone installing it must provide a Tavily API key, and searches or research calls may count against that account's limits or billing.
The skill uses a Tavily bearer API key, which is expected for Tavily API access but authorizes account usage and may consume credits.
Requires TAVILY_API_KEY ... primary_credential: TAVILY_API_KEY
Use a dedicated Tavily key with appropriate limits, store it only in the environment, and do not paste API keys into chat.
Search terms, URLs, and research prompts may leave the local environment and be processed by Tavily.
The artifacts clearly disclose that user queries, target URLs, extraction requests, and research inputs are sent to Tavily's external API.
outbound_hosts: ["api.tavily.com"] ... curl -s "https://api.tavily.com/search" ... -d '{"query":"latest OpenAI API changes"...}'Avoid sending private, confidential, or regulated data in Tavily queries unless that use is acceptable under your data policy.
The agent may run local Python commands to call Tavily and print JSON results.
The skill relies on running a bundled Python CLI. This is purpose-aligned for the API wrapper and not shown as automatic or hidden execution.
Use the terminal tool to run the bundled CLI script ... python3 SKILL_DIR/scripts/tavily.py search --query "latest OpenAI API changes"
Run only the documented commands from the skill directory, review the bundled script if provenance matters, and avoid passing secrets as search text.
Users have less external context for verifying the bundled CLI's origin.
The registry does not show an upstream source or homepage while the skill includes a runnable script. No malicious install behavior is shown, but provenance is limited.
Source: unknown; Homepage: none ... 1 code file(s): scripts/tavily.py
Prefer installing from a trusted publisher or review the script and registry details before use.
A malicious or misleading web page could try to influence the agent's answer if its content is treated as instructions instead of source material.
The skill intentionally brings external web content into the agent's context. Web pages can contain untrusted instructions, even though this is inherent to web search and extraction.
For primary text, run `extract` on 1-3 top URLs ... Use returned `results[].url` fields as citations/sources
Treat extracted web content as untrusted data, verify claims against citations, and do not follow instructions found inside web pages unless the user explicitly asked for them.