The Agent Skills Directory

[SAFE]: The skill is designed with a strong focus on security, implementing clear boundaries for external network access and provider resource management. It uses a structured configuration approach to ensure all search engine interactions are pre-approved and budgeted according to the decision framework.
[COMMAND_EXECUTION]: The skill utilizes a local Python script (scripts/validate_serp_source_config.py) to perform static validation of configuration files. This script verifies that no plain-text secrets, such as API keys or credential-bearing URLs, are present before the configuration is used by subsequent skills.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection attack surface as it handles untrusted data from user keywords and website crawls (ingestion points in SKILL.md and assets/source-config.template.json). While this skill only performs configuration and lacks direct execution capabilities for ingested data, the data is prepared for downstream processing by the harvester skill. No specific sanitization or boundary markers for keyword content are provided in this configuration gate.

serp-source-configurator