The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes data from the Google Search Console API, which serves as a potential vector for indirect prompt injection from external sources.
Ingestion points: API responses returned by the http_json function in scripts/gsc_cli.py.
Boundary markers: The CLI output is raw JSON, providing no specific delimiters or instructions to the agent to disregard embedded content.
Capability inventory: Local script execution, writing to a configuration file, and network access to Google APIs.
Sanitization: The script ensures the response is valid JSON, but does not sanitize text content within that JSON for malicious instructions.
[SAFE]: The skill handles sensitive credentials securely by storing OAuth tokens in ~/.config/gsc-cli/config.json with restricted filesystem permissions (0o600) and masking secrets like access_token and refresh_token in CLI status summaries.
[COMMAND_EXECUTION]: The analytics_query command includes a feature to read local JSON files via the --body-json @path syntax, which is a standard pattern for providing complex input from the local environment.

gsc-cli