dwg-to-excel
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes an external binary named DwgExporter.exe via Python's subprocess.run. The implementation uses a list of arguments rather than a shell string, which is a best practice to mitigate shell injection risks. However, it still grants the agent the capability to execute a local binary with user-influenced file paths.
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to extract text, attributes, and metadata from DWG files. This processed data is then fed back to the agent, creating a vulnerability where malicious instructions embedded in a drawing could attempt to influence the agent's behavior.
- Ingestion points: Data is ingested through the
read_entitiesandget_text_contentmethods in SKILL.md after the CLI tool converts the binary DWG to a readable format. - Boundary markers: The instructions do not define specific delimiters or instructions for the agent to ignore embedded content within the extracted CAD data.
- Capability inventory: The skill possesses
filesystempermissions and the ability to execute subprocesses, which could be targeted by a successful injection. - Sanitization: There is no evidence of sanitization or filtering of the extracted text content (such as annotations or block attributes) before it is presented to the agent.
- [EXTERNAL_DOWNLOADS]: The skill's documentation contains links to external resources, including the vendor's website (datadrivenconstruction.io) and a GitHub repository for the cad2data pipeline. These are legitimate resources provided for documentation and technical support.
Audit Metadata