same-idea-both-platforms
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input, such as ideas or existing posts, to generate drafts for Twitter and LinkedIn. This creates a potential surface where adversarial instructions in the input could attempt to trick the agent into performing unauthorized actions.
- Ingestion points: The skill ingests user content during Step 1 (finding the idea) and when adapting posts between platforms as described in SKILL.md.
- Boundary markers: The instructions do not specify any delimiters or safety markers to isolate user-provided text from the agent's core instructions.
- Capability inventory: The skill utilizes external tools
publish_postandschedule_postwhich perform state-changing actions on social media platforms. - Sanitization: No explicit sanitization, filtering, or validation of the user-provided input is specified.
- [COMMAND_EXECUTION]: The skill references tools that perform external actions, specifically for publishing content to third-party social media services.
- Evidence: The workflow mentions
publish_postandschedule_posttool calls for interacting with Twitter and LinkedIn profiles. - Mitigation: The skill includes explicit security instructions for the agent to never execute these calls autonomously. It requires a "clear yes" from the user and separate approvals for each platform to ensure human oversight.
Audit Metadata