The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's primary function is to instruct the agent to execute the specgate CLI tool. It provides specific command templates such as specgate change status "$WORK_REF" --json, specgate doctor --json, and specgate open. While this involves shell execution, it is the intended purpose of the skill to interface with this specific tool.
[INDIRECT_PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection as it processes data from external sources (repository artifacts and CLI outputs).
Ingestion points: The agent interpolates a user-provided $WORK_REF into a shell command and is instructed to follow instructions contained in the JSON output of specgate change status (specifically data.next_actor and next_command). It also reads "Context Packs" from the repository.
Boundary markers: There are no explicit delimiters or warnings provided to the agent to ignore instructions embedded within the CLI output or repository artifacts.
Capability inventory: The agent has the capability to execute shell commands via the specgate CLI.
Sanitization: The skill does not provide instructions for sanitizing the $WORK_REF input or validating the contents of the next_command field before execution.

specgate-router