linkedin-post-writer
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input and external context data to generate social media content, creating a surface for potential indirect instructions to be executed by the agent.\n
- Ingestion points: Processes user-provided topics, angles, and draft ideas (SKILL.md), and loads data from external references like story-bank.md and voice-profile.md. It also ingests the output of the linkedin-interviewer skill.\n
- Boundary markers: No explicit delimiters or instructions to ignore embedded commands in the ingested data were found in the skill instructions.\n
- Capability inventory: Includes capabilities to publish, update, and delete posts on LinkedIn, as well as fetch external media via media_from_url (SKILL.md).\n
- Sanitization: No explicit sanitization or input validation logic is present to filter malicious instructions from the ingested data before it is processed or used in tool calls.\n- [COMMAND_EXECUTION]: The skill instructions involve calling other platform skills or tools using CLI-like arguments, such as linkedin-humanizer --mode profile and linkedin-interviewer --mode post. These represent the execution of logic based on potentially untrusted input within the agent environment.\n- [EXTERNAL_DOWNLOADS]: The skill can fetch external media from public HTTPS addresses using the media_from_url tool. This is a functional requirement for the skill's purpose of creating media-rich social media posts.
Audit Metadata