The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides a collection of Python scripts in the scripts/ directory (including account_overview.py, create_campaign.py, and upload_audience.py) that the agent is intended to execute in order to interact with the LinkedIn Marketing API for reporting and management tasks.\n- [INDIRECT_PROMPT_INJECTION]: The skill's design facilitates an indirect prompt injection attack surface through its processing of external data.\n
Ingestion points: The agent ingests user-provided content such as campaign names, ad headlines, and file paths (specifically via the --file argument in upload_audience.py) which are subsequently used as parameters for script execution.\n
Boundary markers: There are no explicit delimiters or 'ignore' instructions provided in the skill instructions or script logic to separate untrusted user data from execution logic.\n
Capability inventory: The skill is capable of performing network requests to api.linkedin.com, reading arbitrary local files for audience uploads, and writing configuration data to a local .env file.\n
Sanitization: The included scripts use standard argument parsing but lack explicit sanitization or strict validation of the string inputs before they are transmitted to the external API or used in local operations.

linkedin-ads