The Agent Skills Directory

[SAFE]: The skill consists primarily of documentation and process definitions for ad distribution. It does not contain any executable scripts, binary components, or instructions that trigger automated high-risk operations.
[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for processing external data, such as creative assets in the approved/ directory and campaign metadata in campaigns/<slug>/config.yaml. These ingestion points represent a theoretical attack surface for indirect prompt injection. However, the risk is effectively mitigated by the skill's core rules, which mandate manual human approval for all distribution and spending actions, and rely on an upstream QA gate for initial asset verification. 1. Ingestion points: Campaign configuration files (config.yaml) and approved creative assets in the project directory. 2. Boundary markers: The process requires explicit sign-off from both a QA gate and a human reviewer before distribution begins. 3. Capability inventory: The skill drives distribution through the local oneclick pipeline and the Postiz API/CLI, but all output and spend actions are gated by human verification. 4. Sanitization: Relies on the 'static-qa' and 'vid-qa' stages of the ai-ad-pipeline to filter malicious or non-compliant content before it reaches the distribution stage.

ad-distribution