linkedin

Pass

Audited by Gen Agent Trust Hub on Sep 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses a local Python script, scripts/typefully_scheduler.py, to perform API operations such as listing scheduled posts, listing social sets, and scheduling new drafts. The script is invoked via command line from the main skill workflow.
  • [DATA_EXFILTRATION]: The skill transmits drafted post content and user-provided API keys (via environment variables) to Typefully, a well-known social media scheduling platform (api.typefully.com). This behavior is consistent with the skill's documented purpose.
  • [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection by processing external data from user inputs and file contents to generate LinkedIn posts.
  • Ingestion points: User-provided topics and file paths supplied to the --file argument of the scheduler script.
  • Boundary markers: No specific delimiters or "ignore instructions" warnings are utilized when interpolating data into the post-drafting prompts.
  • Capability inventory: The skill possesses network access (via urllib.request to the Typefully API) and the ability to read local files.
  • Sanitization: There is no evidence of specific input validation or sanitization of the content before it is processed by the agent or sent to the external API.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 23, 2026, 06:07 AM