The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill configures telemetry collection pipelines that ingest external data (logs and traces), which introduces a potential surface for indirect prompt injection.
Ingestion points: The filelog and otlp receivers defined in SKILL.md and references/collector-pipeline.md ingest data from local file systems and network endpoints.
Boundary markers: No specific delimiters or boundary markers are defined for the telemetry content.
Capability inventory: The skill uses the collector to read files and export data to multiple remote backends via network protocols.
Sanitization: The skill provides clear instructions and code to use the attributes processor to hash PII and delete sensitive fields like db.statement or db.connection_string to mitigate data exposure.
[EXTERNAL_DOWNLOADS]: The skill downloads official instrumentation components from the project's public repository.
Evidence: Fetches the opentelemetry-javaagent.jar from the official OpenTelemetry GitHub repository at github.com/open-telemetry.
[COMMAND_EXECUTION]: The skill includes shell commands for environment configuration and agent execution.
Evidence: Uses npm install, pip install, and java -javaagent within the implementation guides in SKILL.md and references/sdk-instrumentation.md.

devops-opentelemetry