openalex
Pass
Audited by Gen Agent Trust Hub on May 5, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads scholarly metadata and PDF files from
api.openalex.organd related Open Access locations. This behavior is consistent with the skill's primary purpose. - [COMMAND_EXECUTION]: Shell scripts (
openalex_query.shandopenalex_download_pdf.sh) usecurlandjqto process data. These scripts correctly use--data-urlencodefor user-provided parameters to prevent command injection into thecurlexecution. - [CREDENTIALS_UNSAFE]: The skill requires an
OPENALEX_API_KEY. It correctly instructs the user to provide this via an environment variable and provides a safe verification command that avoids echoing the secret value to the console. - [DATA_EXPOSURE]: PDF files are downloaded and stored locally. The script
openalex_download_pdf.shincludes a sanitization step (work_id##*/) to prevent basic directory traversal attacks through the work ID parameter. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the OpenAlex API (such as publication titles). While there are no explicit boundary markers to prevent the agent from interpreting this data as instructions, the risk is inherent to any data-retrieval tool and is mitigated by standard agent-side guardrails.
Audit Metadata