fyaic · veil-chow-fyaic · Sep 7, 2026 · Sep 7, 2026
@@ -48,7 +48,9 @@ It needs no Node/MCP/hook setup. [One controlled opted-in desktop pair passed](d
 including prior context, B's own edit and busy/stop checks. A new
 [public-entry checkpoint](docs/09-reviews/2026-09-07-native-public-entry.md) verifies
 anonymous workflow retrieval; title-based activation, novice task selection and
-normal plugin activation remain open. Sending is not race-free. External adapter/hook adoption is
+normal plugin activation remain open. The live readiness attempt returned empty
+current-turn evidence despite host-reported completion; resolve result visibility
+before repeating it. Sending is not race-free. External adapter/hook adoption is
 a separate portability route, not a prerequisite for trying native guidance.
 
 The [desktop-first plan](docs/10-planning/desktop-entry-2026-09-07.md) supersedes

@@ -58,6 +58,11 @@ it. A missing workflow, missing tools or unresolved title is a stopped setup,
 not a successful connection. You can say **Stop ThreadMesh collaboration** in
 each task to cancel it; no separate control panel is required.
 
+If the app shows “completed” but supplies no readable setup confirmation, keep
+collaboration off. Do not assume pairing worked or repeatedly rerun it. This
+occurred in our [readiness attempt](../09-reviews/2026-09-07-native-public-entry.md#authorized-live-readiness-attempt);
+its cause is unresolved, and a completed indicator alone is not acceptance.
+
 This is an explicit workflow request in the existing conversation, **not proof
 that installing a plugin hot-loads old tasks**. The model must actually retrieve
 the workflow, inspect available native tools and confirm the selected scope.

@@ -1,7 +1,7 @@
 # Native desktop public entry: retrieval and readiness boundary
 
-Date: 2026-09-07. Status: **public retrieval checked; name-based desktop pairing
-and activation not yet accepted**. This record does not replace the earlier
+Date: 2026-09-07. Status: **public retrieval checked; live name-based readiness
+attempt not accepted because current-turn evidence is empty**. This record does not replace the earlier
 [controlled native business case](2026-09-07-native-desktop-acceptance.md).
 
 ## What changed for a user
@@ -28,7 +28,7 @@ avoids the inventory; no specific desktop reference-picker UI is asserted.
 | One measured fetch | 910 ms using Node fetch on the maintainer host | Cold-user timing or an onboarding speed guarantee |
 | Skill format validation | Passed the Skill Creator validator in an isolated `uv` dependency environment | Correct model decisions or plugin activation |
 | Repository regression | 459 passed, one optional native test skipped; 55 schema cases and seven transitions passed; 146 Markdown files linted cleanly | Live name resolution or autonomous model behavior |
-| Existing native task name matching | Not run in this checkpoint | A title-only pairing pass |
+| Existing native task name matching | Attempted in both tasks; completed-turn reads returned no items | Successful workflow retrieval, name matching or absence of peer sends |
 | Public-source activation and business edit | Not run in this checkpoint | End-to-end desktop onboarding |
 
 Public workflow revision: `592014782d10a8c4b88f46ea23b7cf588ff78355`.
@@ -42,17 +42,60 @@ then passed using `uv run --with pyyaml python` and the existing validator.
 No user-global skill or plugin was installed. The published alpha.3 archive is
 unchanged; this revised workflow is reached through the pinned source URL.
 
-## Next bounded check
+## Authorized live readiness attempt
 
-The earlier two dedicated native tasks remain stopped. A request was made for
-permission to give each a readiness-only check and one task inventory, without
-reactivating collaboration, sending peer advice or changing sample artifacts.
-No such check has been dispatched at this checkpoint.
+The user subsequently authorized one readiness-only message to each original
+dedicated task and one task inventory per task. Both were `notLoaded` before
+dispatch. The manager explicitly invoked these checks; this was not an automatic
+peer wake or a race-free idle-send test. The messages provided the pinned public
+URL and the other task's exact title, **not its internal ID**, and required a new
+name lookup rather than reusing IDs remembered from the earlier run.
 
-If authorized, record actual public retrieval, uniquely resolved target title,
-available native tools, no outgoing peer message and unchanged business files.
-Do not count already known IDs or a local workflow read as evidence for the
-new entry. This is still a maintainer check, not an independent GUI-user result.
+The narrow prompt required complete public retrieval, at most one authorized
+inventory, actual native tool inspection, and a concise result. It explicitly
+kept collaboration off: no peer messages, business edits, installation, permission
+changes, new tasks or restoration of cancelled advice. The manager made exactly
+two check dispatches and did not retry either model turn.
+
+| Observation | Task A | Task B |
+|---|---|---|
+| Host-reported terminal status | Completed, idle | Completed, idle |
+| Host-reported duration | 142.592 s | 137.989 s |
+| Current-turn items returned by native read | 0 | 0 |
+| Current-turn result/tool evidence | Unavailable | Unavailable |
+| Preceding turn still readable | Yes, six items | Yes, two items |
+
+A bounded reread including the preceding turn reproduced the empty-current-turn
+result. No failure cause is established: a host-reported completion does not
+prove the model ran the requested checks. Do not infer a quota failure, successful
+name resolution, zero peer sends or a confirmed collaboration state from the
+empty history. No enable command was issued; the last readable agreements remain
+stopped. No private host storage or UI automation was used as a workaround.
+
+The three known sample artifacts were unchanged before/after by SHA-256:
+
+| Artifact | SHA-256, identical before and after |
+|---|---|
+| A's `brand.json` | `330315d837dd80782c73fb07ad641458a0259002f4fe0f98e191ccb67c9ed35e` |
+| B's `landing.json` | `267643aca34e3da28f5807f6a49813545c7f849097bc1cf300cf3fee1870a3c7` |
+| B's `price.txt` | `5f9ef93d379ed019937fefa3bada8830ad124e63342acc0aca6da0c0b6ffa38e` |
+
+Native read responses for the current and preceding turn are retained outside
+the repository in owner-only local storage. Their file commitments are:
+A `f8be6a2bc0ba08a8b15ff26190afa40ba39356f89856c3d3c4dbed950e197092`;
+B `81443659b81d849e3cb2f998ec1e7411d5028681a760f04047bd06469715493f`.
+No raw task IDs, histories or inventory contents were published. No screenshot
+or recording exists for this attempt. The read responses are not full execution
+traces and cannot fill the missing tool evidence.
+
+## Disposition
+
+The bounded readiness gate **did not pass**. The bilingual guide now explicitly
+says that a completed indicator without a readable confirmation is not ready.
+Resolve current-turn result visibility before another acceptance run; do not
+paper over it with a new CLI pair, repeated quota-consuming dispatches or a
+claim based on the earlier successful native case. Independent GUI onboarding
+and public-source activation remain open.
 
 Official [skill documentation](https://learn.chatgpt.com/docs/build-skills)
 describes reusable instructions and host loading. It does not itself prove this

@@ -30,7 +30,7 @@ not count as live-product or independent external-verifier evidence.
 
 ## Live attempt audits
 
-- [Native desktop public entry: retrieval checked, title pairing pending](2026-09-07-native-public-entry.md)
+- [Native desktop public entry: retrieval checked, live readiness evidence unavailable](2026-09-07-native-public-entry.md)
 - [Codex-first installed-package acceptance and retained failures](2026-09-07-codex-first-use-release.md)
 - [Codex native desktop: prior context, original receiver edit and busy/stop checks](2026-09-07-native-desktop-acceptance.md)
 - [Codex native-task skill: earlier packaging/tabletop checkpoint](2026-09-07-codex-native-skill.md)

@@ -80,7 +80,7 @@ from the 273-second packaged example and removes the global Node badge: Node is
 a package requirement, not a requirement of the native skill workflow. Native
 attribution and the absence of measured incremental benefit remain explicit.
 
-**Decision: draft only; do not post this turn.** The three requested maintainer
+**Previous checkpoint decision: draft only.** The three requested maintainer
 example/progress/receiver-evidence improvements have shipped. Another README
 rearrangement is not a new functional release. The next useful increment is the
 remaining Codex desktop pairing/activation friction: test a publicly retrievable
@@ -95,10 +95,14 @@ scheduled follow-up is configured by this document.
 Public-entry checkpoint: the bilingual title-and-topic prompt now uses a pinned
 public workflow; anonymous retrieval and byte parity passed. The readiness-only
 mode preserves previous stops. Title matching in the dedicated desktop pair has
-not been dispatched pending scoped permission; do not claim the new entry has
-passed merely because HTTP retrieval did. See the
-[record](../09-reviews/2026-09-07-native-public-entry.md). The draft below remains
-unsent, with no scheduled posting.
+now been attempted with scoped user permission. Both host-reported completed,
+but the current-turn native reads returned no items; preceding turns remained
+readable, and sample artifact hashes were unchanged. This cannot prove title
+matching or zero peer sends. See the
+[record](../09-reviews/2026-09-07-native-public-entry.md). The bounded attempt is
+complete, not passed. After this record merges, post the consolidated released
+Codex example update with this explicit desktop limitation; do not delay the
+contributor's already-delivered priorities behind repeated native retries.
 
 > Thank you, Andrei. We followed the three concrete priorities in your report.
 >
@@ -137,8 +141,8 @@ advantage over native-only use. These are material clarifications to include whe
 a reply is authorized, not reasons to post another immediate acknowledgement.
 
 The user explicitly asked us to make several real improvements before replying
-again. Therefore do not post another acknowledgement, roadmap promise, test
-request or issue closure now. Continue implementation, review and verification
+again. Do not post another acknowledgement-only message, roadmap promise, test
+request or issue closure. Continue implementation, review and verification
 inside the authorized task rather than stopping after each small step.
 
 The next reply should contain the actual released install/run entry, a short

@@ -6,8 +6,11 @@
 title-and-topic prompt with a pinned public workflow. Anonymous retrieval and
 source parity passed; readiness checking is separate from activation. The
 [entry record](../09-reviews/2026-09-07-native-public-entry.md) keeps title-only
-desktop matching and end-to-end activation open. The earlier test pair remains
-stopped; no peer probe or reactivation was dispatched for this checkpoint.
+desktop matching and end-to-end activation open. Two user-authorized read-only
+checks were dispatched: both host-reported completed, but current-turn histories
+were empty while preceding turns were readable. Sample hashes were unchanged;
+readiness and absence of peer sends cannot be verified. No enable command or
+repeat dispatch was issued. Resolve result visibility before another live gate.
 
 **Value and evidence correction:** the native demo uses Codex's own communication
 and continuation; the skill adds guidance, not transport. [English responsibility

@@ -6,6 +6,7 @@ adoption or proof of incremental value over the host's native capability.
 | Recent case | Public record | Retained private originals |
 |---|---|---|
 | Codex desktop prior-context pair | [Actual exchange, B's diff and reduced audit](codex-native-2026-09-07/README.md) | Complete A/B native exports, original tasks and files |
+| Codex public-entry readiness attempt, not accepted | [Empty current-turn evidence and unchanged artifacts](../09-reviews/2026-09-07-native-public-entry.md) | Native read responses retained separately; not full execution traces |
 | Codex default installed-package pass, 272.604 s | [Acceptance](../09-reviews/2026-09-07-codex-first-use-release.md) | Events, report, transcripts and sample artifacts copied to a dated non-temporary archive |
 | Codex diagnostic pass, 184.050 s | [Same record, separate budget](../09-reviews/2026-09-07-codex-first-use-release.md) | Same archive categories; not counted as the default gate |
 | Codex old-runtime failure | [Failure retained alongside passes](../09-reviews/2026-09-07-codex-first-use-release.md#failures-that-changed-the-implementation) | Events and failed-run report copied to the dated archive |

@@ -47,6 +47,10 @@ https://raw.githubusercontent.com/fyaic/threadmesh/592014782d10a8c4b88f46ea23b7c
 流程读不到、工具缺失、名称无法确定，都属于设置未完成，不能当作连接成功。
 要取消，可以分别在两个任务里说“停止 ThreadMesh 协作”，不需要额外控制面板。
 
+如果应用显示“已完成”，却没有可读的设置确认，请保持协作关闭，不要当作配对成功，
+也不要反复重跑。这是我们在[只读就绪检查](../09-reviews/2026-09-07-native-public-entry.md#authorized-live-readiness-attempt)
+中实际遇到的问题，原因尚未确定；“已完成”的状态不能代替验收结果。
+
 这是在已有对话中明确请求使用流程，**不是证明安装插件能热加载旧任务**。
 模型应实际读取流程，核对已有原生工具，并确认选择范围。若读取失败或缺少工具，
 就停在能力检查，不能用新 CLI 会话或私有接口冒充成功。