The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes scripts/run_diagnose.py using the Python interpreter. This script acts as a wrapper for the geo_seo_hub library, which performs the core diagnostic logic.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from user-supplied URLs and HTML files. The security risk is mitigated by a deterministic workflow that uses structural parsing (extracting titles, landmarks, and JSON-LD) rather than raw text interpolation, as well as strict per-source size limits (2 MB) and a limited total fetch budget.
[DYNAMIC_EXECUTION]: The execution script scripts/run_diagnose.py dynamically modifies the Python environment path at runtime to locate internal source code libraries relative to the file's location. This is a standard initialization pattern for the provided diagnostic tool.
[SAFE]: The skill implements significant safety controls for network activity. It enforces explicit public HTTP(S) protocol usage and rejects non-public IP addresses (localhost, loopback, private ranges), which effectively prevents Server-Side Request Forgery (SSRF) attacks. All external fetches are gated by timeouts and size limits.

geo-diagnose