Add Bailey OS Swarm Build Specification

[chrbailey]

Motivation

• Establish a controlled, architecture-first framing for scaling multi-agent work by defining durable work packets and a pod-based organization rather than launching unbounded coding agents.
• Provide a single canonical artifact that captures product doctrine, module boundaries, and the acceptance/critic standards needed to safely hand off work to an execution swarm.
• Make the first-mile build decisions explicit so downstream waves can implement a verifiable spine (CLI, context, claims/evidence, critic/Ralph, reporter) without drifting into premature plugin work.

Description

• Add a new root-level specification file BAILEY_OS_SWARM_BUILD_SPEC.md that includes product definition, primary CLI surface, and core doctrine.
• Define repository layout, architecture backbone, module boundaries, and minimum data schemas for run, claim, evidence, critic results, and Ralph decisions.
• Introduce an Agent Packet Protocol, a 10-pod/100-role map, staged build Waves, acceptance test standards, a Critic-Loop standard, and Ralph routing rules.
• Specify the first milestone bailey audit-repo . and the recommended bailey-os-swarm-brief/ package layout for any large implementation swarm handoff.

Testing

• No automated tests were required for this documentation-only change, as it introduces no runtime code.
• The new file was created and added to the repository for downstream implementation and validation against the specified acceptance tests and critic-loop standards.
• Future implementation PRs should include automated unit/integration tests that exercise the listed acceptance criteria (e.g., bailey init, bailey scan, claim/evidence schema validation, and Ralph routing tests).

---

Codex Task