opencite
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill represents an indirect prompt injection surface because it downloads PDFs from external, potentially untrusted sources and converts them to markdown for the agent to read and synthesize.
- Ingestion points: The
pdf --convertandbatch-fetch --convertcommands (described inSKILL.mdandreferences/cli-reference.md) download academic papers from the internet and convert them to text format which is then provided to the agent's context. - Boundary markers: The instructions lack explicit boundary markers or directions for the agent to ignore instructions embedded within the converted markdown text.
- Capability inventory: The skill has the capability to execute shell commands via
uvx, perform filesystem writes (saving PDFs and reports), and make network requests to various academic APIs. - Sanitization: There is no mention of sanitization or filtering of the converted markdown content before it is processed by the agent.
- [COMMAND_EXECUTION]: The skill relies on the execution of the
openciteCLI tool using theuvxpackage runner for all operations, including searching, downloading, and converting files. - [EXTERNAL_DOWNLOADS]: The skill downloads academic papers from various external sources (Semantic Scholar, OpenAlex, PubMed, arXiv, etc.) and installs the
opencitepackage from a public registry during setup.
Audit Metadata