meta-ads
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to search for and ingest content from local project files named
ad-process.mdandad-brief.mdto customize its recommendations. This ingestion of untrusted local data represents a surface for indirect prompt injection. - Ingestion points: SKILL.md (Instructions to search and read project-specific files).
- Boundary markers: Absent; instructions direct the agent to "read it and apply the user's preferences" without explicit delimiters or safety warnings for the ingested content.
- Capability inventory: The skill allows the agent to execute shell commands through the
adkitCLI tool. - Sanitization: Absent; the agent is instructed to build recommendations based directly on the file contents.
- [COMMAND_EXECUTION]: The skill instructs the agent to check for and utilize a command-line interface tool named
adkit(e.g.,adkit status,adkit meta campaigns create). These commands are used to manage advertising campaigns directly from the terminal if the user has the vendor's software installed. - [EXTERNAL_DOWNLOADS]: The skill references several external domains and repositories for resources and tools.
- Vendor Resources: Links to
adkit.sofor the Ad Library, Creative Studio, and Safe Zone tools. - Educational Resources: References a public GitHub repository (
github.com/coreyhaines31/marketingskills) for lead magnet strategies. - Skill Extensions: Suggests installing an additional skill using
npx skills add adkit-so/ads-skills --skill ad-brief, which fetches content from the author's own collection.
Audit Metadata