Philosophy

Orca exists because agentic software work needs more than speed. It needs a durable definition of success, clean context boundaries, attributable proof, and lifecycle decisions that remain inspectable after the original chat ends.

Working Beliefs

• Good specs make agents faster because they remove hidden interpretation.
• The first blind QA pass is uniquely valuable and cannot be recreated after briefing.
• Subagents should be used for independence, not theater.
• A review gate is useful only when it can block or change the work.
• A project is not done until another capable agent can understand what happened.
• User-facing work should feel product-specific and intentional, not generic, prompt-shaped, or stylistically interchangeable.
• A skill can improve execution, but only the Mission may determine readiness and completion.
• Local state and explicit portable files are preferable to an opaque hosted dependency for the core contract.

Practical Discipline

Orca Mission Control enforces the small contract. The optional workflow library favors artifacts that are short enough to use and specific enough to test. It avoids ceremonial process. Every artifact must help an executor decide, build, verify, review, test, or ship.