engineering-spec

Pass

Audited by Gen Agent Trust Hub on Mar 23, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data to generate technical specifications. 1. Ingestion points: Feature descriptions, design images, and prior skill outputs defined in the Input section of SKILL.md. 2. Boundary markers: Absent; no delimiters are used to isolate user content from instructions. 3. Capability inventory: The skill performs local filesystem writes to telemetry files and suggests invoking external tools like /si (SKILL.md). 4. Sanitization: No input validation or filtering is implemented.
  • [SAFE]: The telemetry logging functionality targeting ~/.claude/projects/ is a localized operational feature for skill improvement within the platform's configuration space and does not represent a malicious data exfiltration or exposure risk.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 23, 2026, 10:58 AM
Security Audit — agent-trust-hub — engineering-spec