The Agent Skills Directory

[COMMAND_EXECUTION]: The skill triggers shell commands via 'omc autoresearch' and orchestrates other execution skills ('omc-plan', 'autopilot', 'ralph', 'team') after requirements are finalized. This is the intended behavior for bridging requirement analysis to implementation.
[PROMPT_INJECTION]: The skill processes user-supplied ideas and codebase information which are interpolated into evaluation prompts, creating a surface for indirect prompt injection. * Ingestion points: User arguments and local files read by the 'explore' agent. * Boundary markers: Not explicitly defined in the prompt templates used for ambiguity scoring or specification crystallization. * Capability inventory: Includes the ability to write files ('Write'), invoke other skills ('Skill'), and execute specific shell commands. * Sanitization: No specific sanitization or filtering logic is described for the content ingested from external sources before its use in generation steps.

deep-interview