The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill identifies runtimes and entrypoints by scanning project metadata files, creating a vulnerability surface for indirect instructions embedded in developer files.
Ingestion points: Project files including pom.xml, requirements.txt, and package.json are read by otel_runtime_scan.go.
Boundary markers: The documentation does not specify the use of delimiters or 'ignore' instructions when parsing these files.
Capability inventory: The skill possesses file-writing capabilities (otel_collector.go), network access for ingestion monitoring (ingest_watch.go), and subprocess execution (otel_process.go).
Sanitization: The instructions do not detail any sanitization or validation procedures for the ingested file content.
[EXTERNAL_DOWNLOADS]: The skill resolves and downloads the OpenTelemetry Collector binary and Java agent JAR from GitHub.
Evidence: otel_collector.go identifies the latest release on GitHub and downloads the appropriate binary for the host platform (darwin/linux/windows). This is documented as vendor-provided functionality.
[COMMAND_EXECUTION]: The skill manages the execution of external binaries and instrumented applications through a controlled subprocess wrapper.
Evidence: otel_process.go utilizes StartManagedProcess() to launch subprocesses and monitor their lifecycle, including PID tracking and port detection.

dtwiz-otel