feat(observation): qualify Linux Codex client-ID delivery observation #130

Current roadmap — 2026-10-13 KST

• Target milestone: 0.5.0 — Cross-platform observation and remote qualification.
• Planning priority: P2 (priority:P2).
• Dependencies/order: feat(codex): add a shared bounded app-server client #71/feat(codex): add opt-in delivery observation by client message ID #72 after 0.4.0. feat(observation): qualify Windows Codex handoff storage and delivery observation #131 also needs private Windows ACL/storage and owned-process-tree qualification. Linux/Windows tracks are independent.
• Remaining scope: Exact platform/native profiles and client-ID injection evidence. Other-platform, queue-only or fixture evidence cannot qualify native observation.

This section supersedes older milestone/version/priority statements below. Existing acceptance criteria, evidence limits and historical records remain applicable. Versioned milestones are provisional planning targets, not committed dates or implementation/live-test/merge/release authorization. P1 means a milestone prerequisite/current-release item; P2 a planned extension/research track; P3 later dependent work.

Scope and priority

P2 follow-up, Future — Agent integrations. User approved keeping #69–#72 together in 0.4.0 with explicitly qualified macOS arm64 observation; Linux observation is a later platform extension, not a new 0.4.0 release blocker. No implementation, live experiment or release is performed by registering this issue.

The 0.4.0 candidate PR #109 qualifies Codex 0.160.1/0.162.1 on Darwin arm64. Actual KR/US Linux arm64 tests established ordinary SSH queue acceptance, and one US operator-confirmed receipt; they did not qualify Linux client-ID injection observation. Do not infer support from Online, matching SemVer, a schema, visible response or Darwin fixtures.

Acceptance criteria

• Independently pin exact Linux OS/architecture/native version/source/schema/artifact profiles and implement platform-aware profile selection, initialization, owned-PID storage binding and owner inspection. No version-range or cross-platform inheritance.
• Preserve exact original home/thread/client ID/generation and retained qualification during send and independent wait. Read-only bounded metadata projection; no transcript/rollout API, warm-up, model turn, wake or extra writer. Document whether SQLite internally accesses body-bearing cells; never promise a stronger privacy boundary than proven.
• Preserve total monotonic deadlines and owned-process cleanup, private ledger permissions/fencing and at-most-one queue attempt. Required unsupported goals refuse pre-effect; best-effort fallback is allowed only before an app-server attempt; no resend after uncertainty.
• Deterministic fixtures for storage/config binding, stale/changing/wrong owner, schema/version drift, injection vs queue deletion/turn completion, timeout/SIGINT, restart/resumed wait and complete/partial SSH results.
• With separate live-test authorization, demonstrate one exact client-ID injection in a dedicated thread on every claimed profile; ordinary KR/US queue results are historical evidence only.
• Update capability/support matrix, API/PARITY and all locales; queued/delivered are not receipt ACK or work completion.

Related #71/#72, PR #109. Relay/MCP, generic wake, receiver dedup and automatic ACK collection are outside scope.