bear-support

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the scimaster-cli tool (via the sci command) to check versions, initialize authentication, and perform academic searches. These operations are required for the skill's primary functionality of retrieving research data.
  • [EXTERNAL_DOWNLOADS]: The skill instructions direct the user to install the scimaster-cli package using npm install -g scimaster-cli. This is the official installation method for the tool required by the skill.
  • [PROMPT_INJECTION]: The skill processes untrusted external content (paper abstracts and metadata) retrieved through the CLI, which constitutes an indirect prompt injection surface. Ingestion points: Metadata and abstracts from sci search results. Boundary markers: Absent; there are no specific delimiters to protect against malicious instructions embedded in abstracts. Capability inventory: The skill has the capability to execute CLI commands and write multiple report files to the local directory. Sanitization: Content is translated and formatted into reports, providing a basic layer of transformation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 01:39 PM
Security Audit — agent-trust-hub — bear-support