langchain-agent-monitoring
Fail
Audited by Snyk on Nov 20, 2026
Risk Level: CRITICAL
Full Analysis
CRITICAL E004: Prompt injection detected in skill instructions.
- Potential prompt injection detected (medium risk: 0.30). The skill is a legitimate LangSmith monitoring guide, but setting
is_tracing_disabled: truekeeps the judge's own runs out of the user's trace volume, which is a mild concealment of the skill's activity, so I classed it as medium under the recall-first override.
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (medium risk: 0.30). The required workflow runs a decision-model evaluator over production agent threads (
human_ai_pairsfrom live traces), so end users of the deployed agent can inject free text into a feed the workflow reads continuously without the agent selecting any item.
MEDIUM W013: Attempt to modify system services in skill instructions.
- Attempt to modify system services in skill instructions detected (low risk: 0.10). The skill creates, patches, and deletes online evaluator rules, decision models, and Insights configs on a production tracing project, which changes the monitoring environment, but it gives no explicit unrecoverable loss of operational state, so the irreversible-loss gate fails and the tier is low.
Issues (3)
E004
CRITICALPrompt injection detected in skill instructions.
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
W013
MEDIUMAttempt to modify system services in skill instructions.
Audit Metadata