The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the bash tool, specifically calling a local script scripts/agent/merge-queue.sh. This is the intended primary purpose of the skill for managing the merge queue.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data in the form of task identifiers (e.g., SH-XXXX) and branch names during the merge process. This creates a surface for indirect prompt injection if the inputs are malicious.
Ingestion points: Task identifiers and branch names passed as arguments to the merge commands (e.g., scripts/agent/merge-queue.sh merge --task SH-XXXX).
Boundary markers: The skill documentation does not define boundary markers or explicit instructions for the agent to ignore instructions embedded in branch names or task metadata.
Capability inventory: The skill uses bash to execute shell scripts, grep for searching logs, and read for file access. These capabilities could be abused if the script processing the task ID is vulnerable to command injection.
Sanitization: There is no evidence in the instructions of sanitization or validation for the task identifiers or branch names before they are interpolated into shell commands.

merge-queue