linkedin-content
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill points to external installation instructions hosted on a GitHub repository (raw.githubusercontent.com/inference-sh/skills).
- [REMOTE_CODE_EXECUTION]: Instructions encourage the installation of third-party command-line utilities (belt-sh/cli) and platform-specific skills to perform tasks such as image generation and automated posting.
- [INDIRECT_PROMPT_INJECTION]: The skill templates ingest user-provided content for LinkedIn posts and pass it as input to external AI models and search assistants.
- Ingestion points: User content is interpolated into JSON payloads for the 'belt' command-line interface in SKILL.md.
- Boundary markers: No explicit delimiters are used to separate user-provided text from command parameters or system instructions.
- Capability inventory: The skill utilizes command execution via the 'belt' tool and network-based app invocation.
- Sanitization: Example commands do not demonstrate any escaping or validation of user-supplied strings before they are processed by external tools.
Audit Metadata