cpu-profile-analysis

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [Indirect Prompt Injection Surface]: The skill is designed to ingest and parse untrusted profile and trace data files, which creates a potential exposure to indirect prompt injection if those files contain crafted content.
  • Ingestion Points: In SKILL.md, external data enters the context when user-provided .cpuprofile and Trace-*.json files are loaded via fs.readFileSync.
  • Boundary Markers: Absent. The instructions do not define specific delimiters or instructions to isolate the parsed content or prevent the model from interpreting text embedded within fields like function names or URLs as instructions.
  • Capability Inventory: The provided scripts demonstrate file-system capabilities, specifically file-write operations using fs.writeFileSync to reformat profile data.
  • Sanitization: Absent. The data is parsed structurally using JSON.parse or buffer operations, but individual string values (such as function identifiers) are not sanitized or escaped before being processed or presented in final reports.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 10:13 PM
Security Audit — agent-trust-hub — cpu-profile-analysis