alpha-research

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute specific commands using the alpha CLI tool via a bash shell. These commands (search, get, ask, code, annotate) are well-defined and scoped to the skill's purpose of research paper analysis.
  • [PROMPT_INJECTION]: The skill processes untrusted external data from academic papers and GitHub repositories, creating a surface for indirect prompt injection.
  • Ingestion points: Data enters the agent context through search results and full-text content retrieved via alpha search, alpha get, alpha ask, and alpha code.
  • Boundary markers: There are no explicit delimiters or instructions provided to the agent to separate the retrieved external data from its internal instructions.
  • Capability inventory: The agent interacts with the alpha CLI through bash commands to perform research tasks.
  • Sanitization: No explicit sanitization or filtering of the fetched external content is documented in the skill instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 12:47 AM