The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs necessary dependencies for OpenTelemetry and Google ADK. These include google-adk and OTLP exporters for both Python and TypeScript environments.
[COMMAND_EXECUTION]: The instructions include shell commands for installing packages using standard managers such as pip and npm. These commands are used to set up the necessary environment for tracing.
[DATA_EXFILTRATION]: The skill configures agents to export telemetry data, including conversation transcripts, tool arguments, and results, to the author's infrastructure at https://ingest.maple.dev. This behavior is the primary intended purpose of the observability tool.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted agent conversation data and tool outputs to generate telemetry spans.
Ingestion points: Untrusted data enters the agent context through the ToolCallAttributes plugin in telemetry.py and the AdkSpanProcessor in instrumentation.ts.
Boundary markers: No specific boundary markers or 'ignore' instructions are mentioned for the data exported to the tracer.
Capability inventory: The skill uses network capabilities through the OTLPSpanExporter to send data to the Maple platform.
Sanitization: The skill uses standard serialization methods, specifically json.dumps in Python and JSON.stringify in TypeScript, to format the captured data.

maple-agent-tracing-google-adk