edgartools
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- EXTERNAL_DOWNLOADS (MEDIUM): The skill requires the installation of the
edgartoolspackage from PyPI. While it is a known library, it is not from a specifically trusted source list provided in the security guidelines. - PROMPT_INJECTION (HIGH): Indirect Prompt Injection Surface: The skill downloads and processes external data from the SEC EDGAR API which could contain malicious instructions targeting the agent.
- Ingestion points: Data enters the agent context through
filing.text(),filing.markdown(), andfiling.reportsinSKILL.md. - Boundary markers: There are no markers or instructions to the agent to ignore embedded instructions within the filings.
- Capability inventory: The skill demonstrates local file-write capability to
/tmp/and the ability to execute code using the installededgartoolspackage. - Sanitization: No evidence of sanitization or validation of the fetched filing content before it is processed or written to disk.
- COMMAND_EXECUTION (MEDIUM): The skill utilizes the
uvpackage manager to execute theaddcommand and performs Python file-write operations to the local filesystem at/tmp/. This can be exploited if the ingested data contains malicious content that influences the agent to perform unauthorized file operations.
Recommendations
- AI detected serious security threats
Audit Metadata