fix(ci): restore full-public catalog contracts

[JasonBuildAI]

Impact: The daily 514-check full-public-smokes release readback cannot report ready for release owners and every release candidate: two deterministic catalog checks fail on current main, forcing verification to waive them and masking the signal of the other 512 passing checks.

Goal And Delivered Outcome

• Outcome basis / optional anchor: issue [Bug]: daily full-public fleet remains not ready because two public catalog contracts are stale #6794; follow-up to fix(ci): restore full-public fleet readiness #6740 in the same full-public fleet-readiness class.
• Goal/source and gap: the latest scheduled run (39068476762) reports ready=false with these two failures, and both still fail on main at a54c030ee (the catalog smoke on Linux and Windows, the configure-goal smoke on Linux). IP-030's machine-configuration inventory omits the registered codex_chat namespace, and the goal-configuration catalog smoke pins a feature set without decision_context.
• Observable before → after, with the validation row that proves it: both public smokes exit 1 at a54c030ee (catalog: missing ['codex_chat']; configure-goal: the set(features) assertion at line 290); both exit 0 at be7736408. The regression_parity row records the before/after.
• Issue/task and intended base: Fixes [Bug]: daily full-public fleet remains not ready because two public catalog contracts are stale #6794; base loopx-project/loopx main.

Author Declaration

• Written by: AI agent (DeepSeek Harness, deepseek-flash) at @JasonBuildAI's request, direction and review.

Implemented against

• Specification and revision: issue [Bug]: daily full-public fleet remains not ready because two public catalog contracts are stale #6794 and the release-readback rule in docs/development/testing-and-quality.md at a54c030ee; the current catalog owners are loopx/configuration_catalog.py and loopx/capabilities/native_chat/machine_defaults.py.
• Criteria:

Criterion (spec clause)	Disposition	Symbol / path	Test or command
IP-030 inventory lists every registered machine-configuration namespace	implemented	docs/concepts/interaction-pattern-catalog.md:1214	python examples/interaction-pattern-catalog-smoke.py
Goal-configuration smoke pins the current catalog feature set, including decision_context	implemented	examples/project/configure-goal-smoke.py:290	python examples/project/configure-goal-smoke.py
No product behavior change; only public expectations and docs	implemented	diff	git diff --check and both smokes

• Self-check before submission: I read both failing assertions at a54c030ee, reproduced both on Linux, applied only the two expectation updates, and re-ran both smokes at be7736408. I did not run the scheduled full-public workflow; its next daily or manual run is the end-to-end confirmation.

Scope And Continuation

• Completed scope and remaining work: complete within this scope. Both known red checks are fixed, so the next release-candidate full readback can be read without waiving them.
• Slice boundary / successor: N/A. The next full-public run confirms ready=true; other concurrent fleet work is separate.

Validation

• Tested revision: be7736408f244918bed6402f8bcbdb715ec5aafb (the rebased head bf544980b1cb8a6525a3985b5b127bbdb8f784d6 carries the byte-identical two-file patch, git blob 60dce0123249bb393f18bdc441e40fc2631f41b3; the catalog smoke re-ran green at the rebased head)
• Run state: finished
• Input classes: synthetic

Check kind	Result	Public-safe evidence / limitation
regression_parity	passed	At base a54c030ee both smokes exit 1: the catalog smoke reports missing ['codex_chat']; the configure-goal smoke raises the feature-set assertion. At head be7736408 both exit 0; the rebased head bf544980b carries the byte-identical patch and the catalog smoke re-ran green there.
integration	passed	python examples/interaction-pattern-catalog-smoke.py -> interaction-pattern-catalog-smoke: ok (native Windows, head be7736408).
integration	passed	python examples/project/configure-goal-smoke.py -> configure-goal-smoke ok (Linux Python 3.12 container with Node v24.22.0, head be7736408).
static	passed	git diff --check; no whitespace errors.
static	passed	ruff check examples/project/configure-goal-smoke.py and python -m py_compile examples/project/configure-goal-smoke.py both pass; loopx check --scan-path docs/concepts/interaction-pattern-catalog.md --scan-path examples/project/configure-goal-smoke.py reports a clean public boundary scan (2 files, errors=0, warnings=0).
integration	failed	loopx canary premerge --from-git-diff --git-diff-base upstream/main: all 5 direct checks pass and 16 of 17 selected checks pass. The only failure is content-ops-issue-fix-metadata-preview-smoke.py, which passes standalone and is unrelated to the two changed files; it fails under the premerge's isolated HOME (known host issue).

• Coverage and gaps: the two changed files are the complete public contracts behind the two failing checks. The scheduled full-public-smokes workflow does not run on PRs, so ready=true is confirmed by the next daily or manual run; this PR does not claim that run. Required PR checks (Sign-off, merge-gate) were still queued at independent-review time; no failure was observed. The local premerge failure above is an unrelated environment/runner issue, not a failure of the changed contracts.

Frontend / Visual Evidence

• UI impact: none
• Before: N/A
• After: N/A
• First viewport and preview: N/A; documentation-only inventory text and a public smoke expectation, with no user-visible UI surface.
• States and viewports shown: N/A
• Source data: none
• Attention review: N/A; no UI change.
• Craft review: N/A; no UI change.

Type of Change

• Bug fix
• New feature
• Breaking change
• Refactoring (no functional changes)
• Documentation update
• Test update

LoopX Area

• Control plane (goals, todos, quota, scheduler, registry, runtime)
• Benchmark boundary (adapters, runners, verifiers, scoring, evidence)
• Capability or extension (providers, adapters, skills)
• Public docs or presentation surface (README, protocols, dashboard)
• Build, packaging, installer, or CI
• Host or runtime integration

Technical Direction

• Direction / acceptance reference, when applicable: N/A

Shared-authority RFC fixture impact

• Production-scale fixture schema: N/A
• Semantic dimensions changed, or reviewed no-impact rationale: N/A
• Provider conformance arms run: N/A
• Read-only legacy/file/PostgreSQL three-arm rehearsal (required for promotion, runtime-routing, or compatibility-projection changes): N/A

Boundary Checklist

• Neither the diff nor this PR body/comments/attachments disclose private state, credentials, raw traces or verifier output, internal links, or local machine paths (including .loopx/, .codex/goals/, and live ACTIVE_GOAL_STATE.md).
• I did not duplicate maintainer-owned benchmark work unless a maintainer split out a public issue for it.
• I kept the change scoped to the linked issue/task.
• Existing English/Chinese RFC pairs (including ledger entries) both have a diff, and the local premerge check python3 scripts/check_rfc_bilingual.py --base origin/main --head HEAD passes, or no paired RFC changed.
• I completed the visual evidence section for UI changes, or marked UI impact none.
• Every commit includes a DCO Signed-off-by trailer (git commit -s).

AI disclosure: this change was drafted with AI assistance at my request, under my direction and review.

[loopx-agent]

Reviewer: model_agent · gpt-6.1-sol · OpenAI · runtime_reported · reasoning_effort=max

APPROVE：两处契约漂移已独立复现并验证修复，未发现阻断项。 Exact head: 6801@bf544980b1cb8a6525a3985b5b127bbdb8f784d6；base: 860e42d4b63b45c4798ee7f08622a27c146a2610。全量 fleet 与质量凭据资格分开判断。

动机

维护者需要通过公开目录和配置 smoke 分辨真实回归，而不是每次都绕过已经登记的新能力。

原目录遗漏已注册的 codex_chat，配置 smoke 又遗漏已有 decision_context，两条检查在当前基线失败；补齐后两条原检查均通过，原有严格集合和默认关闭边界仍保留。

本机同一解释器下，两条命令分别从基线失败变为当前通过；目录说明和原生配置描述一致，不修改产品输出。

本 PR 只修复两处公开契约漂移，不改变产品行为，也不宣称整套每日 smoke、发布候选或托管 CI 已经健康。

改动思路

现有产品 owner 已经提供两个描述，问题在公开契约没有随之更新。因此只扩展原 IP-030 目录与原 configure-goal smoke，不新增测试脚本、第二份配置注册或运行时兼容分支。目录仍由原生注册表校验完整性；配置测试仍使用精确 feature 集合，并额外检查 decision_context 只是本地预览，没有 default 或 current 激活状态。

不需要新增开关或让使用者再配置一次；这是恢复既有检查的信号，而非用新预期覆盖错误产品输出。旧失败对应当前已接受并存在的 namespace/descriptor，已分别核对其生产 owner。

具体改动

关键内容讲解

完整差异仅 2 个路径、11 行增加和 1 行删除：

1. IP-030 补上 codex_chat，并说明它只存普通新 Goal 会话的 profile alias、model、effort、tier，不改变已有对话身份，不管理 steward/attached Agent/Agent Loop，不存凭据或授权。
2. 既有 configure-goal smoke 补上 decision_context，再断言 availability=local_artifact_preview 且不存在 default/current。原有其他 feature、精确集合和各项设置断言完整保留。

规范来源 #6794，spec_revision=2026-10-18T08:13:49Z，以及基线 测试与全量 fleet 边界。映射标签 IP-030.codex_chat 和 configuration_catalog.decision_context 均 implemented。核对的现有生产 owner 是 native_chat machine_defaults.py、machine configuration builtins、configuration_catalog.py 及 decision_context/catalog_entry.py，不是从通过后的输出临时编造预期。

对主干的风险

风险主要是把不合法产品输出“测试成合法”，或把发现/预览写成启用。这次没有修改任何生产路径：命名空间确已注册，Decision Context 的既有 descriptor 本来就只有预览能力而无 default/current。精确集合没有改成包含关系，新增断言也强化了预览边界；无权限、凭据、状态存储或新默认行为变化。

在实际不可变基线 860e42d4 上，原目录 smoke 报 missing codex_chat，原配置 smoke 报 feature-set 断言；在当前 head 两个原命令均退出 0。解释器与真实临时 CLI 路径相同，没有复制成功输出或仅测试 parser。当前标准 premerge 5 direct＋17 selected 全部通过，公开边界扫描干净；托管总体结果仍因 stale quality receipt 不通过，需原生质量流程重新确认当前范围。

作者旧 revision 曾报告 isolated-HOME 下 metadata-preview 失败；保留该历史，本轮当前 exact head 的相同被选检查通过，没有声称找到旧环境根因或修好了间歇问题。没有查询、触发、等待 CI，也没有运行整个每日/发布 fleet；两条检查通过不等于完整候选 ready=true。同作者当前三项提案分别处理恢复、目录契约和 quota 缺状态，主题不同；既有 #6740 修的是另外两条 vision/backup smoke，本次没有重复创建维护者工作。

我的整体评价

这个有界仓库维护目标可判 goal_achieved；long_horizon=improved：移除已复现的契约噪声，保留对后续真实漂移的严格检查；user_experience=preserved：产品旅程和用户准备工作不变，维护者读回更清楚。

有界可维护性检查选择继续扩展原目录与原 smoke，没有新增框架或近似注册表，当前无需额外重构。这是有价值的契约维护，不以测试数或修复历史红灯宣称新产品能力。批准仅对此 exact head；过期质量凭据、完整 fleet 和维护者合并授权保持独立。

English verdict: APPROVE at bf544980b1cb8a6525a3985b5b127bbdb8f784d6. Both original public smokes fail at the immutable base and pass here. Existing product owners support the added inventory and strict preview-only assertions; no runtime behavior changes. Full-fleet health and stale exact-scope quality qualification remain separate.