The Agent Skills Directory

[SAFE]: The skill provides legitimate instructions for implementing trace assertions using the OpenTelemetry SDK. It demonstrates how to use the InMemorySpanExporter and SimpleSpanProcessor to capture and verify spans in a test environment.
[EXTERNAL_DOWNLOADS]: The skill describes installing standard packages from well-known registries, such as opentelemetry-sdk via pip and @opentelemetry/sdk-trace-base via npm. These references target official OpenTelemetry libraries, which are widely recognized and trusted services.
[INDIRECT_PROMPT_INJECTION]: The skill describes an attack surface where the agent processes trace data (spans and attributes) that could theoretically contain untrusted content if the application under test is processing external requests.
- Ingestion points: Trace spans are ingested through the InMemorySpanExporter (e.g., memory_exporter.get_finished_spans() in SKILL.md).
- Boundary markers: No specific delimiters are used for the span attributes within the instructions.
- Capability inventory: The skill uses standard assertions (e.g., assert in Python, expect in TS, assertThat in Java) to verify the data.
- Sanitization: No explicit sanitization of span attributes is mentioned. However, because this is performed within a controlled test environment using a local in-memory buffer, the risk is negligible.

opentelemetry-trace-assertions