signals-scout-ai-observability

Pass

Audited by Gen Agent Trust Hub on Jun 15, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted LLM trace data which creates an attack surface for indirect prompt injection. An attacker could potentially embed instructions in application traces intended to manipulate the scout's findings or poison its long-term memory.
  • Ingestion points: Telemetry data containing raw request/response pairs is retrieved using query-llm-trace (documented in SKILL.md and referenced in references/lenses.md).
  • Boundary markers: None. The skill does not use delimiters or instructions to treat trace content as data rather than instructions when processing telemetry events.
  • Capability inventory: The skill can persist findings to a durable scratchpad via signals-scout-scratchpad-remember and post to a user-facing inbox via signals-scout-emit-signal (SKILL.md).
  • Sanitization: No validation or sanitization is performed on the retrieved trace content before it is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 15, 2026, 10:58 AM
Security Audit — agent-trust-hub — signals-scout-ai-observability