[purpshell]

Summary

Reconcile campaign A/B and weighted-variation human output on accepted CLI main 044f8b75ae8b0689664fe489c22d4a3c64486717. Published head: 7ad623a4876582aaa1cb80b92f87156c90105497, tree 449f1eaae905a1368d112025e03eb6eab1a1758b. The guarded rewrite replaces old afc9e016813211ee0af9b9d6e44421a966cfd2e3, preserving its author/date, helper and all 13 test cases.

The additive reconciliation preserves current follow-up, drip and frequency rendering, create flags/run body, agent setup and browser-shell changes. It adds escaped A/B outcomes and per-variant analytics, weighted counts, unknown-shape refusal and supported beta help. README guidance describes the existing update command correctly. JSON output and transport remain unchanged.

Exactly five source paths plus genuine contracts/cli-reference.json output change. Producer scripts, browser registry/policy, package manifest/lock and SDK pin remain byte-identical to 044f.

Impact

API behavior and feature policy are unchanged. Help claims were checked against accepted API eda77760b56095a771c55e5d2df18be23e985377. SDK source and package acceptance are separate; the CLI pin is unchanged. CLI human output/help and generated reference change here. Docs bundle adoption, registry publication, feature activation and environment acceptance remain held. Admin actions and controls are unchanged.

Verification

The completed local command, before this publication sweep, used a private copy of an existing lock-matching dependency graph with Node 24.15.0. No installer or shared donor write was used.

• Unchanged 13 cases against compiled 044f: 4 controls pass, 9 expected missing-feature failures. Combined-source successor: all 13 pass with explicit OCLIF_TEST_ROOT.
• Format, lint, typecheck, supported build, genuine reference generation/check/tests, browser registry generation, actual emitted-browser verifier and offline package dry-run pass.
• Browser registry and support policy stay byte-identical to 044f; emitted assets are qualification artifacts only. No hardcoded command count or manual reference edit.
• SDK pin and locked graph remain unchanged. Full suite, installed package/environment acceptance and hosted new-head CI are not claimed. Historical failures and resource receipts are retained. No broad tests, CI rerun or review was added for publication.

Fresh Git remote reads before guarded push still returned main 044f and expected old PR head afc9. A connector snapshot reported another base; it was not used to change source or claim a new integration.

Session provenance and UI evidence

Same Campaigns source owner, thread 646a0c0a-e900-4d0f-9a25-0b8834d7d4e6, Codex gpt-6.1-sol. Original campaign checkpoint: committed session evidence. The refreshed source/command/dependency/backup receipts are retained in the owner's publication packet. CLI fixture output only; no graphical UI or live provider acceptance.

Rollout and rollback

Keep draft. Release alone sequences merges. Revert the feature commit to remove the display/help additions; no state or schema changes. Matching published SDK/package and API environment evidence remain required for release. CLI72/79 stay held at their existing heads and pins. No registry publishing, feature activation, catalog date change or staging deployment.

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

_{Comment @coderabbitai help to get the list of available commands.}