The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the gpc CLI, such as gpc train start, gpc train status, and gpc train abort. These commands are used to manage the release train lifecycle and are legitimate for the skill's purpose.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from a local configuration file, .gpcrc.json, which defines release stages and gates. Although this is an ingestion point for user-controlled data, it represents standard configuration for a developer tool.
Ingestion points: .gpcrc.json in the project root and ~/.cache/gpc/train-*.json state file.
Boundary markers: None present; the skill expects structured JSON data.
Capability inventory: Execution of gpc train shell commands for release management.
Sanitization: No explicit validation or sanitization of configuration content is mentioned.

gpc-train