Stabilize Every Code parity on Codex CLI fork baseline

[shiny-code-bot]

Finish Line

Current main is treated as a provisional Codex CLI fork baseline until Every Code feature concepts and regression fixtures have been reconciled, with must-preserve overlays validated by fixtures before broad implementation ports.

Current Status

State: Active parity recovery. Current main is a provisional Codex CLI fork baseline, not full Every Code parity.

Landed:

• PR docs: add Codex fork parity ledger #402 added docs/codex-fork-parity-ledger.md as the fixture/test/probe view for Stabilize Every Code parity on Codex CLI fork baseline #397.
• PR docs: add Codex fork parity ledger #402 marked current main as provisional in the feature inventory and upstream policy docs.
• PR feat: restore prompt-cache hit-rate diagnostics #403 restored the first prompt-cache hit-rate diagnostics slice for Audit token and prompt-cache diagnostics parity #401.
• Make code-exec-harness the Dogfood Parity 1 exec gate #405 / PR Make exec harness a dogfood gate #406 established tools/code-exec-harness as the deterministic no-live-token P0 code exec --json gate against the checkout-built code binary.
• Add exec-harness safety scenarios for Dogfood Parity 1 #407 / PR Add exec harness safety scenarios #408 added deterministic exec safety coverage for shell command round-trip behavior and read-only write denial.
• PR Unblock build-fast from native WebRTC metadata fetch #409 closed Port build-fast cache mode from stale build-cache branch #396 by removing native WebRTC git dependency metadata fetches from the default local build graph; local ./build-fast.sh is usable again.
• Dogfood Parity 1: git and patch mutation harness safety #410 / PR Add git patch exec harness safety scenarios #411 added deterministic git/patch mutation safety coverage: workspace file-content assertions, git-status assertions, apply_patch round-trip/file-change/tool-result evidence, workspace-write edit persistence, and read-only denied-file absence.
• Dogfood Parity 1: install binary replacement confidence #412 / PR Add local code rebuild preflight check #413 added install/binary replacement confidence: non-mutating local rebuild preflight, deterministic temp-fixture checks, explicit symlink rollback messaging, and guarded PATH code --version verification only when PATH resolves to the rebuilt binary.
• Child parity issues exist for Restore Auto Drive parity fixtures on Codex fork baseline #398 Auto Drive, Define Code Bridge and browser control parity boundary #399 Code Bridge/browser, Map Auto Review proof metrics onto Codex review primitives #400 Auto Review proof metrics, and Audit token and prompt-cache diagnostics parity #401 token/prompt-cache diagnostics.

Realignment:

• Dogfood Parity 1: Daily Driver Baseline #404 is the Dogfood Parity 1 daily-driver baseline under this parent.
• Keep using current origin/main as the provisional Codex CLI fork baseline; do not resume broad work from spike/codex-base-port.
• Native WebRTC is intentionally outside the default build graph for now; restore it later only as an explicit opt-in overlay so fresh checkout metadata remains reliable.

Current rule:

• Continue from current origin/main, not spike/codex-base-port.
• Treat missing Every Code behavior as unclassified until the ledger or a linked issue records Port, Rewrite, Covered, Defer, or Retire.
• Port fixtures/tests/concepts before implementation overlays where feasible.
• Do not claim daily-driver parity until Dogfood Parity 1: Daily Driver Baseline #404 acceptance criteria are met.
• Keep every code-bearing parity slice behind ./build-fast.sh, just harness-smoke when relevant, and scoped review evidence.

Next action:

• Create/land the next Dogfood Parity 1: Daily Driver Baseline #404 child slice for one high-value overlay behavior. Prefer Auto Review proof metrics (Map Auto Review proof metrics onto Codex review primitives #400) if the goal is actionable review evidence and less broad-review noise, or Code Bridge/browser (Define Code Bridge and browser control parity boundary #399) if the goal is interactive app debugging parity.
• Then continue overlay fixtures for Auto Drive, Code Bridge/browser, Auto Review proof metrics, and token/burn diagnostics one concept at a time.

Acceptance Criteria

• Inventory old Every Code fixtures, snapshots, integration tests, and behavior probes from fa0c33944f / pre-feat: replace Rust substrate with Codex base #390 tree against current origin/main.
• Build a mechanical deleted-test ledger, especially under code-rs/tui/tests, code-rs/tui/src/**/snapshots, code-rs/core/tests, app-server protocol fixtures, bridge/browser tests, review tests, and Auto Drive tests.
• Classify each fixture/behavior as one of: port as-is, rewrite against Codex-base primitive, covered by new Codex test, required Every Code overlay, replace with Codex-native behavior, defer, or retire with rationale.
• Create focused follow-up issues for must-preserve overlay domains before implementation starts.
• Port must-preserve fixtures/tests first when feasible, or document why a fixture must follow implementation.
• Keep Desktop/app-server startup compatibility from Validate Codex-base Every Code in copied Codex Desktop app #387 green while parity work proceeds.
• Keep Decide remote inbox and Discord UI for Codex-base session continuity #388 remote-inbox/session-continuity gates intact until external-session discovery/resume/approval evidence exists.
• Add or update docs/release guardrails so release-bound work does not imply product parity before the ledger is resolved.
• Update Inventory and port Every Code features onto Codex base #386 with the resulting overlay classifications.
• Only after parity-critical fixtures/concepts are captured should spike/codex-base-port be deleted as historical evidence.

Initial Domains

• Auto Drive.
• Code Bridge and browser/control tooling.
• Multi-agent workflows and agent telemetry.
• Auto Review and proof metrics.
• Token, rate-limit, prompt-cache, and burn diagnostics.
• Remote inbox and externally-created session continuity.
• GitHub/LaunchPlane/cloud automation.
• Config compatibility and CODE_HOME behavior.
• TUI approval/session/resume/rollback behavior.
• Build, release, local dogfooding, and cleanup workflows.

Relationships

• Parent strategy: Keep efficiency work aligned with just-every/code and Codex CLI #51.
• Substrate/policy: Plan Codex-base substrate migration for Every Code #384 and Realign docs around Codex CLI fork plus Every Code overlays #395.
• Feature inventory: Inventory and port Every Code features onto Codex base #386.
• Desktop validation: Validate Codex-base Every Code in copied Codex Desktop app #387.
• Remote-inbox/session-continuity gate: Decide remote inbox and Discord UI for Codex-base session continuity #388.
• Build-cache/build-fast blocker: Port build-fast cache mode from stale build-cache branch #396.
• Dogfood Parity 1: Dogfood Parity 1: Daily Driver Baseline #404.
• related: Keep efficiency work aligned with just-every/code and Codex CLI #51 - Keep efficiency work aligned with just-every/code and Codex CLI #51
• related: Plan Codex-base substrate migration for Every Code #384 - Plan Codex-base substrate migration for Every Code #384
• related: Realign docs around Codex CLI fork plus Every Code overlays #395 - Realign docs around Codex CLI fork plus Every Code overlays #395
• related: Validate Codex-base Every Code in copied Codex Desktop app #387 - Validate Codex-base Every Code in copied Codex Desktop app #387
• related: Decide remote inbox and Discord UI for Codex-base session continuity #388 - Decide remote inbox and Discord UI for Codex-base session continuity #388
• related: Port build-fast cache mode from stale build-cache branch #396 - Port build-fast cache mode from stale build-cache branch #396
• related: Dogfood Parity 1: Daily Driver Baseline #404 - Dogfood Parity 1: Daily Driver Baseline #404

[cbusillo]

Parity stabilization progress: #403 merged (6ff1ffa) as the first #401 token diagnostics slice.

What changed:

• Added a visible prompt-cache hit-rate signal on current Codex-base token plumbing without restoring old token/account modules wholesale.
• Added prompt-cache-hit-rate to the configurable footer status line.
• Documented current coverage and remaining telemetry/schema/rate-limit gaps in docs/token-diagnostics-parity.md.
• Updated docs/codex-fork-parity-ledger.md to link the Audit token and prompt-cache diagnostics parity #401 diagnostic note.

Validation:

• focused tests: cargo test -p codex-tui prompt_cache
• snapshot refresh/check: INSTA_UPDATE=always cargo test -p codex-tui status_snapshot
• required gate: ./build-fast.sh passed, with the existing unlocked-build/Cargo.lock warning.
• PR blob-size policy passed.

Review note: repository background auto-review continues to fail before findings because the matching-head diff scope exceeds its configured size limit (1,479,472 chars > 120,000). A focused read-only review agent reviewed #403's small diff, found stale snapshots and release-note overclaiming, and both were fixed before merge.

[cbusillo]

Ledger update: #398 closed via #428 by retiring standalone Auto Drive.

The parity ledger now treats the deleted Auto Drive coordinator/TUI/crate surface as retired, not pending rebuild. Future work should preserve only goal-mode-compatible concepts on current goal/review/session/agent primitives. This avoids carrying a large fork-only coordinator diff while keeping the useful regression ideas available as focused goal-mode fixtures.

[cbusillo]

Plan update landed in #429.

Clarification now recorded in repo docs:

• Codex CLI is the upstream substrate.
• Every Code is layered on top as product behavior, defaults, release policy, and additive integrations.
• code-rs is the editable Every Code product workspace built on that Codex CLI substrate.
• codex-rs remains the read-only openai/codex:main mirror for upstream reference/provenance.
• Future alignment should first bring code-rs closer to current Codex CLI shapes, then add Every Code overlays where they are still desired.

#429 also removed active stale Auto Drive leftovers: README /auto promises, the Auto Drive memory watcher LaunchAgent/script, and the exec harness auto scenario wiring to --auto. Historical/retirement docs remain only where they explicitly mark standalone Auto Drive as retired.