The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch and process external HTML content from user-provided URLs. This creates a surface for indirect prompt injection where malicious instructions hidden in a website's metadata or JSON-LD blocks could be processed by the agent.
Ingestion points: Untrusted content is ingested via WebFetch or the fetch_page.py script from external URLs.
Boundary markers: The instructions do not define specific delimiters or warn the agent to treat the fetched content as untrusted data.
Capability inventory: The skill has access to powerful tools including Bash, Write, and WebFetch (as defined in allowed-tools).
Sanitization: There are no instructions for sanitizing or filtering the fetched content before the agent parses it.
[COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python script located at ~/.claude/skills/geo/scripts/fetch_page.py. Because this script is not bundled with the skill (as noted in the compatibility section), its contents and security posture are unverifiable from the skill's source alone.

geo-schema