XGJK BP Audit

Key things to check before installing or enabling this skill: - Credentials: The bundled script expects an API key (BP_OPEN_API_APP_KEY) but the skill metadata declares no required env vars; docs also mention XG_BIZ_API_KEY. Confirm which env var the deployer must set, and ensure the skill metadata/registry is updated to declare that primary credential before installation. - Endpoint and scope: The script defaults to a production base URL (https://sg-al-cwork-web.mediportal.com.cn/open-api). Verify that this is the intended backend for your environment. If you want to test safely, point BP_OPEN_API_BASE_URL to a non-production test endpoint. - Write operations: The CLI supports add_key_result and add_action (POST). Ensure you only allow the skill to perform writes after explicit user confirmation. Prefer disabling autonomous invocation for this skill or require an explicit confirmation interaction when a write action is requested. - Path and runtime consistency: SKILL.md example call paths ('.cursor/skills-v3/...') do not match the actual script location ('scripts/bp-audit/bp_api.py'). Update documentation or the path used by the agent so runtime commands won't fail or cause accidental full-tree searches. - Dependencies: The script uses the Python requests library but the package doesn't declare dependencies. Ensure the runtime has required Python libraries or add a requirements file to the skill manifest. - Least privilege: Provide the skill with an API key that has only the minimal permissions needed (prefer read-only for audits unless user explicitly requests a write), and rotate/revoke keys used for testing. - Code review: If you rely on this skill in a sensitive environment, review the full bp_api.py script for any hidden endpoints, logging of sensitive headers, or unexpected network destinations. Although no obfuscated code was found, the source is from an unknown origin — confirm provenance with the publisher if possible. If you cannot confirm or correct the above (declare primary credential, fix env-var name mismatch, update paths, and ensure explicit user confirmation on writes), treat this skill as suspicious and avoid granting it access to production credentials or enabling autonomous invocation.