Bootstrap formal specification infrastructure, agent skills, and codebase reference docs

[ahouseholder]

Summary

The project lacked a machine-readable, AI-navigable specification layer. Decisions were captured ad hoc in notes, ADRs, and code. This issue tracks the work to establish the scaffolding needed to track requirements, validate them programmatically, and export them for consumption by coding agents.

Deliverables

Spec Infrastructure (specs/, src/ssvc/metadata/specs/)

• YAML specification files covering API, CI, codegen, domain model, namespaces, registry, testing, and versioning
• specs/spec-registry.yaml — master registry of all spec files
• Python package with schema.py, registry.py, lint.py, llm_export.py, render.py
• Unit tests under src/test/metadata/specs/

Agent Skills (.agents/skills/)

• load-specs — flat JSON export for coding agents
• study-project-docs — loads context before implementation work
• process-concerns — batch-processes CONCERNS.md into GitHub Issues
• ingest-idea — converts GitHub Idea issues into formal specs and implementation notes

Codebase Reference Docs (docs/reference/codebase/)

• ARCHITECTURE.md, CONCERNS.md, CONVENTIONS.md, INTEGRATIONS.md, STACK.md, STRUCTURE.md, TESTING.md

Notes (notes/)

• namespaces.md, terminology.md, versioning-rules.md

Packaging Fix

• pyproject.toml: set setuptools packages.find where=["src"] so the ssvc package is importable when installed

Resolved by

PR #1142