The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using oma agent spawn to start various specialist agents (PM, backend, frontend, etc.) in background processes with separate workspaces.
[INDIRECT_PROMPT_INJECTION]: The skill monitors workflow progress by reading human-readable artifacts such as progress and result files (e.g., progress-{agentId}-{taskId}-{runId}-{sessionId}.md). Because these files are generated by other agents that may ingest untrusted input from the codebase or users, there is a risk that malicious instructions could be embedded in these reports and subsequently interpreted by the coordination agent.
Ingestion points: Markdown progress and result files located in the path defined by {memoryConfig.basePath}.
Boundary markers: The instructions do not specify delimiters or "ignore embedded instructions" warnings for the agent when reading these status files.
Capability inventory: The skill has the capability to spawn new agent processes and manage file system workspaces.
Sanitization: There is no mention of sanitizing or validating the content of the progress files before the agent acts upon the status information.

oma-coordination