The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the backlog and worker reports which could influence foreman behavior. Ingestion points: Reads task definitions from .moai/state/kanban/backlog.json and worker completion reports from .moai/reports/<card-id>/evidence.md. Boundary markers: Absent. Task text and evidence paths are interpolated into sub-agent prompts without explicit delimiters or instructions to ignore embedded instructions. Capability inventory: Uses Bash for git and moai CLI operations, and the Agent tool to spawn sub-agents with write access to isolated worktrees. Sanitization: Absent. The skill relies on the operator's manual 'pick' action as a security gate but does not perform content validation.
[DYNAMIC_EXECUTION]: The skill instructs the agent to generate and run a persistent shell script to monitor the backlog file. The script uses a while true loop with cksum and sleep 5 to detect file changes. While the script logic is static and performs only local integrity checks, it constitutes runtime script generation and execution.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute git commands (status, log, diff) and a vendor-specific CLI tool (moai todo). These operations are scoped to the project repository and are consistent with the skill's workflow management purpose.

moai-kanban-foreman