upstash-redis-kv

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it retrieves data from an external Redis store and possesses capabilities to modify or delete data.
  • Ingestion points: Data retrieved from Upstash Redis via operations like GET, HGETALL, and SMEMBERS in scripts/upstash-client.ts.
  • Boundary markers: SKILL.md defines a confirmation protocol for destructive actions, but allows bypassing it via YOLO mode instructions which could be triggered by injected text.
  • Capability inventory: High-impact operations including data deletion (DEL), database wiping (FLUSHALL), and record modification in scripts/upstash-client.ts.
  • Sanitization: Data retrieved from Redis is not sanitized or escaped to prevent the agent from interpreting it as instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 06:05 PM
Security Audit — agent-trust-hub — upstash-redis-kv