The Agent Skills Directory

[Prompt Injection] (LOW): The skill is susceptible to indirect prompt injection. It retrieves and processes untrusted content from the web which could contain malicious instructions designed to manipulate the AI agent's behavior.
Ingestion points: Web search result snippets and raw page content from the Tavily API.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are provided in the skill definition.
Capability inventory: The skill can execute local shell scripts and perform network operations.
Sanitization: No sanitization of the external web content is mentioned.
[Data Exposure & Exfiltration] (LOW): The skill performs network requests to api.tavily.com. While necessary for functionality, this domain is not included in the trusted whitelist.
[Command Execution] (LOW): The documentation describes the execution of a local script ./scripts/search.sh. Although this is a standard operational pattern, it represents a command execution vector.
[No Code] (SAFE): Only the SKILL.md documentation was provided. The referenced bash script (search.sh) was not available for direct security auditing.

search