cybersec-review

Pass

Audited by Gen Agent Trust Hub on Aug 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Analysis of external software repositories presents a potential attack surface for malicious instructions embedded in source code or documentation.
  • Ingestion points: The agent processes repository files, including code, configuration, and documentation (SKILL.md).
  • Boundary markers: The instructions provide behavioral constraints but lack technical delimiters to isolate untrusted repository data from the agent context.
  • Capability inventory: The skill authorizes execution of local scanners, linters, and package-manager audit commands (SKILL.md).
  • Sanitization: The workflow requires manual command inspection and redaction of sensitive findings before report generation.
  • [COMMAND_EXECUTION]: The skill uses local CLI tools for automated vulnerability scanning and code analysis.
  • Evidence: The reviewer is instructed to use 'existing linters or static analyzers' and 'lockfile and package-manager audit commands' (SKILL.md).
  • Mitigation: The skill enforces strict boundaries, including requiring explicit user permission for network access and mandatory inspection of project-invoked scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 18, 2026, 10:14 PM
Security Audit — agent-trust-hub — cybersec-review