algorithm-design-planner

Pass

Audited by Gen Agent Trust Hub on May 5, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [SAFE]: No malicious patterns, prompt injections, or persistence mechanisms were detected. The skill's instructions focus on structured document generation and research planning.
  • [EXTERNAL_DOWNLOADS]: The skill employs WebSearch and WebFetch tools to verify the novelty of research ideas. These operations target public literature and represent standard research workflows.
  • [COMMAND_EXECUTION]: Although the Bash tool is listed in allowed-tools, the instructions do not contain any specific commands for execution, script downloads, or environment modification.
  • [PROMPT_INJECTION]: Analysis of indirect prompt injection surface (Category 8): 1. Ingestion points: --idea argument (SKILL.md) and WebFetch results; 2. Boundary markers: Absent; 3. Capability inventory: Bash, Write, and Edit tools are available (SKILL.md); 4. Sanitization: Absent. This flags a potential surface for instructions in external data, though the risk is assessed as safe given the skill's planning-focused purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
May 5, 2026, 12:40 PM