capture

Pass

Audited by Gen Agent Trust Hub on Aug 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill performs command execution by calling the gbrain CLI utility with user-supplied data as arguments or through standard input.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests arbitrary content from users or external sources and stores it in a queryable 'brain' database, creating an attack surface for indirect prompt injection.
  • Ingestion points: Untrusted data enters the system through gbrain capture using inline strings, --file paths, or --stdin as defined in SKILL.md.
  • Boundary markers: The documentation does not specify the use of delimiters or 'ignore' instructions to prevent the agent from accidentally executing instructions contained within the captured data.
  • Capability inventory: The skill has the capability to execute shell commands (gbrain) and write to the file system (inbox/*) as defined in the YAML frontmatter and body of SKILL.md.
  • Sanitization: There are no documented steps for sanitizing, escaping, or validating the ingested content before it is committed to the database.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 29, 2026, 06:18 PM
Security Audit — agent-trust-hub — capture