bear-onboard
Pass
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of an external Node.js package
scimaster-clivia npm (npm install -g scimaster-cli) to perform its core functions. - [COMMAND_EXECUTION]: The skill executes shell commands using the
sciCLI tool, includingsci --version,sci init, andsci searchwith various parameters to retrieve research data. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted external data (academic abstracts) returned by the search tool and reformats them for the final report.
- Ingestion points: Data enters the context via the
abstractfields in the JSON results ofsci searchcommands. - Boundary markers: None identified; search results are integrated directly into the markdown and HTML report generation process.
- Capability inventory: The agent can execute shell commands via the
scitool and perform file system writes to create the{topic-slug}/directory and associated report files. - Sanitization: No specific sanitization or filtering of the retrieved abstract content is mentioned before it is translated or reformatted into the report.
Audit Metadata