luxonis-project-interview

Pass

Audited by Gen Agent Trust Hub on Jul 7, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches project examples and technical documentation from official vendor repositories (github.com/luxonis/oak-examples) and documentation endpoints (docs.luxonis.com).
  • [COMMAND_EXECUTION]: Optionally executes shell commands to run vendor-provided examples and installs associated dependencies in isolated environments. This capability is gated by user approval and requires a verified device environment.
  • [PROMPT_INJECTION]: The skill ingests untrusted data from the local filesystem which constitutes an indirect prompt injection surface. * Ingestion points: Reads project manifests, README.md, and source files from the local working directory (SKILL.md Section 1). * Boundary markers: Instructions do not specify delimiters or guidelines to ignore embedded instructions in ingested files. * Capability inventory: Performs file writes to PROJECT_BRIEF.md and executes shell commands during the optional example run phase (SKILL.md Sections 5 and 7). * Sanitization: No validation or sanitization is performed on the content of ingested local files before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 7, 2026, 08:07 AM
Security Audit — agent-trust-hub — luxonis-project-interview