serilog

Pass

Audited by Gen Agent Trust Hub on Aug 17, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill documents standard usage and best practices for Serilog, a widely-used structured logging library for .NET. The patterns shown align with industry standards for application diagnostics and observability.
  • [INDIRECT_PROMPT_INJECTION]: The skill demonstrates logging data from external sources, which is a common pattern that introduces an attack surface for log injection.
  • Ingestion points: External data is ingested from HTTP request headers (User-Agent) and user identifiers (ClaimTypes.NameIdentifier) in the request logging middleware examples within SKILL.md.
  • Boundary markers: The instructions explicitly advocate for message templates ({PropertyName}) over string interpolation to ensure data is treated as structured properties rather than being interpreted as part of the log message format.
  • Capability inventory: The skill configures file-system writes (WriteTo.File) and network-based log sinks (WriteTo.Seq, WriteTo.OpenTelemetry, WriteTo.Elasticsearch) for log transport.
  • Sanitization: The skill provides specific guidance on avoiding sensitive data (passwords, tokens) in logs and implementing destructuring limits (ToMaximumDepth, ToMaximumStringLength) to prevent memory exhaustion or large-scale data exposure from deep object graphs.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 17, 2026, 03:54 PM
Security Audit — agent-trust-hub — serilog