The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the hubspot CLI and jq to perform CRM operations. These operations (search, list, get, update) are standard for administrative and marketing automation tasks. The skill provides clear instructions on using these tools safely, including the use of --dry-run for bulk updates.
[DATA_EXFILTRATION]: While the skill processes sensitive PII (contact emails, names, and job titles), all data remains within the local environment (saved to .jsonl files) or within the HubSpot ecosystem via the CLI. There are no network requests to unauthorized third-party domains or suspicious exfiltration patterns detected.
[CREDENTIALS_UNSAFE]: The skill does not contain any hardcoded API keys or secrets. It uses a placeholder value (12345) for owner IDs and directs users to utilize the authenticated hubspot CLI, which is a secure practice.
[PROMPT_INJECTION]: The instructions are strictly functional and intended to guide the agent in data management. There are no attempts to bypass safety filters or override system instructions.
[SAFE]: The skill architecture is consistent with its stated purpose of audience targeting and CRM data segmentation. It follows logical vendor patterns for the 'hubspot' author and leverages official tooling.

audience-targeting