The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell-based repository inspection using Grep, Glob, and Bash through a dedicated 'Explore' subagent to gather codebase facts.
[COMMAND_EXECUTION]: The skill performs file system write operations to save planning artifacts and state in the .omw/plan/ directory.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks as it ingests untrusted content from the codebase to inform the planning process.
Ingestion points: Codebase content retrieved via repository inspection tools in the SKILL.md execution policy.
Boundary markers: The instructions lack explicit delimitation or 'ignore embedded instructions' warnings for data retrieved from the codebase.
Capability inventory: The agent has access to shell execution, file writing, and the ability to spawn multiple specialized sub-agents (Planner, Architect, Critic, Analyst).
Sanitization: No sanitization or validation protocols are defined for content retrieved from external files before it is processed by the planning agents.

plan