[OrcaWin]

Current disposition: deferred for this RAM project; keep draft. YAML remains loaded through required startup config work, so no individual full-idle saving was demonstrated. Evidence #33912 is merged independently into main; its tooling/report provides no reason to merge this implementation.

ELI5

Main-process modules can refer to Orca YAML contracts without loading the parser immediately.

What Changed

Extract the existing parser/normalization into a dependency-injected shared module. Keep the renderer/CLI facade on its existing eager ESM path, and use a main-only synchronous first-use YAML loader for parsing/editing, Hermes config, update feeds, model config, and theme parsing. Preserve text/alias/merge limits and recipe constants.

Why

This separates main-only loading policy from shared browser/CLI behavior. Fresh-profile Hermes setup still legitimately parses YAML during startup; there is no demonstrated full-idle YAML saving in that workload.

Visual Proof

N/A — main-process loading/lifecycle changes; no rendered UI change. Hidden Playwright CDP checks on the complete stack passed, including the rendered SSH add-host/test/refusal flow; see the evidence PR.

Testing

452 relevant parser/Hermes/update/model tests passed (1 skipped); Node typecheck and independent parser/limit parity review passed. Integrated Linux build passed. The combined recommended gate passed 23/24 checks (22,643 tests passed, 3 failed). The three daemon initialization failures also reproduce at the pinned upstream base; the complete gate remains failing. See #33912 for the control and exact cases.

Local runtime evidence is Linux x64/Electron 43.8.1. macOS physical footprint and native Windows/macOS behavior are not measured locally. Remote, folder workspace, mobile and serve contracts remain in scope; there is no wire-protocol change.

Review

Historical experimental stack 4/6. Depends on #33894. Review this PR against its configured base branch.

Evidence and benchmark: #33912. The historical measured stack measures −4.44 MB median idle V8 heap and −5.87 MB median Linux RSS (n=3, 120 seconds, fresh hidden profiles, telemetry/GPU disabled). macOS physical footprint is unverified. M3 is a separate experiment on performance hold; M4/M5 do not make xterm/YAML cold at full idle.

Agent skill upstream boundary

• Not applicable; no upstream skill-installer content copied.

Checklist

• Scope and rationale explained; self-reviewed for correctness, security and performance.
• Cross-platform, remote/SSH, mobile, serve and folder workspaces considered.
• Final validation results and memory report attached in perf(bench): measure startup memory and cold SSH latency #33912.