The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands like find, grep, and test to detect the presence of specific platform markers (e.g., AndroidManifest.xml, Package.swift, package.json). These commands are used for classification and routing purposes.
[COMMAND_EXECUTION]: Probes are implemented with security best practices, including a maxdepth limit of 4 and the exclusion of sensitive or bulky directories such as .git, node_modules, build, and Pods.
[PROMPT_INJECTION]: The ForgeRock detection probe scans file contents for specific strings, which represents a surface for indirect prompt injection. However, the risk is mitigated by the use of the -l flag, which ensures that only matching filenames are returned to the agent, preventing the ingestion of potentially malicious content from the files themselves.

ping-orchestration-sdk-router