The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to run local scripts for lifecycle hooks and task verification, such as ./scripts/board-drained.sh. These scripts are used to evaluate the state of the task board and determine if the loop should continue.
[INDIRECT_PROMPT_INJECTION]: The orchestrator reads and acts upon task states and feedback stored in a shared task list file, which is authored and updated by multiple agents in the team.
Ingestion points: The shared task list file described in references/agent-teams.md is the primary source of truth for the loop.
Boundary markers: The instructions do not define specific delimiters or instructions for the agent to ignore potentially malicious content within task descriptions or feedback stored in the shared file.
Capability inventory: The lead agent is configured with powerful tools including Bash, Write, Edit, Agent, and Task in the SKILL.md frontmatter.
Sanitization: The skill uses jq --arg for safe JSON construction in its shell scripts, but does not specify sanitization for the natural language feedback processed by the agent when evaluating task completion gates.

orchestrator-task-loop