skills/sjunepark/custom-skills/teach/Gen Agent Trust Hub

teach

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No security issues or malicious patterns were identified in the skill logic or metadata.
  • [PROMPT_INJECTION]: The skill instructions are focused on pedagogical outcomes and do not contain patterns intended to bypass AI safety filters or override system-level instructions.
  • [COMMAND_EXECUTION]: There are no shell commands, subprocess invocations, or system modifications defined within the skill.
  • [DATA_EXFILTRATION]: No network communication or external data transfer mechanisms are present.
  • [PROMPT_INJECTION]: The skill ingests codebase files to provide explanations, creating a theoretical indirect injection surface. However, the risk is negligible as the skill lacks any high-privilege capabilities such as file writing or network access. 1. Ingestion points: Reads source code files (SKILL.md). 2. Boundary markers: Absent (snippets are embedded directly). 3. Capability inventory: None (text output only). 4. Sanitization: Absent.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 04:51 AM