The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install OpenTelemetry and OpenInference instrumentation packages via pip and npm. These include openinference-instrumentation-langchain, opentelemetry-sdk, and their TypeScript equivalents. These are well-known, standard libraries for AI observability.
[COMMAND_EXECUTION]: The skill provides a Python one-liner in Step 0 to check the versions of installed packages (python -c "import langchain..."). This is a benign diagnostic command.
[DATA_EXFILTRATION]: The skill's primary function is to capture and export agent conversation data, including transcripts and tool results, to the vendor's platform (ingest.maple.dev or ingest.eu.maple.dev). This behavior is transparently documented as the skill's purpose and uses the vendor's own infrastructure.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests untrusted agent conversation data (input.messages) to generate telemetry.
Ingestion points: Conversation messages are processed by the OTel instrumentation in tracing.py and genai-spans.ts.
Boundary markers: None specific to the message content, as the skill treats them as data to be logged.
Capability inventory: The skill uses OTLPSpanExporter to send data to the network.
Sanitization: The skill supports redaction features (hide_inputs, hide_outputs) provided by the openinference-instrumentation library to prevent sensitive data exposure.

maple-agent-tracing-langchain