linkedin-year-audit
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external LinkedIn analytics Excel files. The agent extracts "hooks" from post URLs to categorize content into themes, creating a surface where a maliciously crafted post title could attempt to influence the agent's behavior during the analysis phase.\n
- Ingestion points: The
scripts/linkedin_year_audit.pyscript reads the user-supplied.xlsxexport file.\n - Boundary markers: While the instructions specify the process is local, they lack explicit delimiters or warnings for the agent to ignore instructions embedded within the extracted hook text.\n
- Capability inventory: The skill can execute a local Python script, read user files, and write findings to markdown and JSON files. It does not have network access.\n
- Sanitization: The script uses
unicodedata.normalizeand regular expressions to clean and normalize the extracted text from URLs.\n- [COMMAND_EXECUTION]: The skill instructs the user to run a local Python script (scripts/linkedin_year_audit.py) to process the analytics export. This is a standard and necessary part of the skill's functionality, performed entirely within the user's local environment.
Audit Metadata