Require production evidence before adoption

VERSION

@@ -1 +1 @@
-0.1.17
+0.1.18

adlc/factory-implement/SKILL.md

@@ -3,8 +3,8 @@ name: factory-implement
 description: Implement accepted application work in its designated workspace and produce a working candidate with evidence for independent review.
 license: MIT
 metadata:
-  version: "0.1.15"
-  updated: "2026-10-15"
+  version: "0.1.18"
+  updated: "2026-10-17"
 ---
 
 # Implement

adlc/factory-implement/references/review-handoff.md

@@ -1,6 +1,6 @@
 # Make the change easy to review
 
-Revision: 0.1.15 · Updated: 2026-10-15
+Revision: 0.1.18 · Updated: 2026-10-17
 
 Use when preparing or revising a PR or substantive handoff for a person. The
 implementer owns this presentation; independent review checks its claims, and
@@ -23,6 +23,22 @@ Return the handoff itself, without narrating compliance with this method. Combin
 problem, example and checks in a short paragraph when sufficient; do not add a
 separate heading for every fact or repeat the same limitation in several places.
 
+Connect the promised outcome to the artifact actually produced and served. A test
+that substitutes its own resource list (such as a `cfg(test)` stand-in), fixture or
+proxy output, a successful build or a hash of the input tree does not show that a
+packaged, generated or embedded resource reached production. When the change
+affects one, inspect the consumed build output and use a supported non-test
+production path on that exact artifact, comparing it with the payload the source
+should produce. Scale this to the change, and catch a missing resource before an
+owner discovers it. For example, if the source should embed eight editor assets,
+show eight in the built package and the editor loading through its normal route,
+not six passing tests. For an artifact going to adoption, this proof comes before
+adoption is authorized or performed, and missing or failed proof holds adoption. A
+development build allowed to omit the bundle does not need it. If no supported
+runtime exposure exists, say which bytes are provenance-only or unverified. Name
+the native acceptance still open after authorized adoption and who can perform it;
+installed does not mean accepted.
+
 ## Choose evidence that explains this change
 
 Start with what the reviewer needs to understand about this specific change.

adlc/factory-review/SKILL.md

@@ -3,8 +3,8 @@ name: factory-review
 description: Independently assess a concrete candidate against its accepted scope, project standards and actual verification evidence before delivery.
 license: MIT
 metadata:
-  version: "0.1.13"
-  updated: "2026-10-12"
+  version: "0.1.18"
+  updated: "2026-10-17"
 ---
 
 # Review
@@ -40,6 +40,14 @@ material failures. Check whether the format explains the actual change and the
 reader's question; flag decorative or redundant visuals. Clearly distinguish
 explanatory diagrams from observed behavior.
 
+Check that each claimed path and result corresponds to the inspected candidate and
+the artifact it actually consumes or serves. Check that the evidence could have
+revealed a realistic omission or failure. A test that substitutes its own
+resource list does not show what the production build contains. Do not demand every possible
+test or invent a universal source-to-runtime check. Absent native behavior is a
+defect only against an established expected contract; otherwise report the open
+question.
+
 Consider consequences and affected trust/data boundaries, not just diff size.
 Use a scoped security review when needed. Compare before/after evidence for claims
 that require it. Report a missing baseline or unavailable environment honestly.

agent-ops/factory-agent-ops/SKILL.md

@@ -3,8 +3,8 @@ name: factory-agent-ops
 description: Act as a project's lead for ideas, research, accepted tasks or an approved backlog; coordinate native workers, independent review and authorized delivery.
 license: MIT
 metadata:
-  version: "0.1.17"
-  updated: "2026-10-15"
+  version: "0.1.18"
+  updated: "2026-10-17"
 ---
 
 # Factory Agent Ops
@@ -61,6 +61,23 @@ authorized check output. Read their evidence and resolve
 findings; a successful tool call or finished turn does not establish acceptance.
 Recheck affected evidence after fixes, rebases or changed scope.
 
+Before reporting delivery, reconcile the children's actual completed outcomes and
+findings with every agreed acceptance condition for the delivered artifact, kept in
+the existing issue or handoff. Carry accepted owner amendments into the current
+writer and review criteria; do not silently narrow the task to the latest fix.
+When an artifact going to adoption affects expected packaged, generated or
+embedded resources, require the applicable proof on that exact artifact, covering
+its consumed output and a supported non-test production path, before authorizing
+or performing adoption. Missing or failed proof holds adoption under any existing
+delegation; native workflow acceptance can follow authorized adoption.
+An acknowledged handoff or a green CI aggregate does not by itself establish a
+condition it did not test, or complete the promised workflow; keep the evidence it
+does provide, and keep BLOCKED or NOT RUN conditions open. Run the checks you can.
+For a genuinely host-only check, name the next holder able to perform it. Keep labels
+such as source-qualified or installed-but-unaccepted until native acceptance
+passes, and do not ask the owner to repeat basic packaging checks. Absent native
+behavior is not a defect until an established expected contract makes it one.
+
 Write issue openings, progress updates and final handoffs for the person reading
 them, in English unless the user explicitly requests another language; chatting
 in another language is not such a request. Lead with

docs/qualification.md

@@ -1,6 +1,6 @@
 # Qualification
 
-Revision: 0.1.17 · Updated: 2026-10-15
+Revision: 0.1.18 · Updated: 2026-10-17
 
 This record distinguishes source review, disposable exercises and observed native
 operation. A package release does not qualify another host, every future update
@@ -10,6 +10,7 @@ remain with the installation; public issues track the remaining acceptance work.
 | Boundary | Observed evidence | Limit / next check |
 | --- | --- | --- |
 | Package | Eight portable skills, metadata/link checks and staging tests | Mechanical checks do not establish agent decisions |
+| Production artifact evidence (0.1.18 candidate) | Reconciled 17 October 2026 from product-lead handbacks ([Factory #76](https://github.com/arcitai/factory/issues/76); not reproduced here). OPD: the source expected eight embedded editor assets, but the installed executable held six and served editor-not-built. Its six tests substituted a `#[cfg(test)]` `tests::BUNDLE_FILES` list, while production uses `OUT_DIR/bundle.rs`, so those tests did not exercise production inclusion. The OPD lead's original handoff left a production smoke optional; the corrected R10 proof gate made it required. A same-source replacement was atomically adopted at 06:17:35Z (transferred receipt). AIOS: the contrast build installed at 06:18:55Z (298 files). Bridge: the owner passed queue history. The lead checked retained artifacts offline against the eight source-derived names: the old six-asset output was rejected for missing `editor.js` and `panel.js`, the fresh eight passed, and the R10 manifest and receipt agree. Six of the eight assets are runtime bytes; two are build provenance | Offline reuse of historical evidence; no binary was run, and this does not show that the method caused the recovery. Still open: the OPD native workflow and app-origin binding, and the AIOS native colours and opening. The missing Bridge toast is not a proven defect; no custom Undo is authorized. In one fresh-context rehearsal of five incident packets, every decision kept these limits: the incomplete package, the installed but unaccepted workflow and visual checks, the toast without an established contract, and the dropped accepted centre amendment. This tested decisions only, with no live product action and no claim of general reliability. Independent review of the exact candidate and its checks is recorded in the delivery PR |
 | Blocked-request follow-through (0.1.17 candidate) | 2026-10-15, after the held OPD request recorded below ([Factory #32](https://github.com/arcitai/factory/issues/32); coordinator native UI actions plus Factory native readback). Declining the held compound shell call failed it without running it, but the run stayed active and proposed a second shell change before the prepared text-only correction arrived. The coordinator then cancelled that held call: neither call ran, the run was interrupted and ended with nothing pending, and its pre-recovery checkpoint was kept. The OPD lead sent its accepted native-file-only instruction to the same worker at that checkpoint; the next run completed native file changes with no shell or approval items and nothing pending while still active | Operator-assisted progress after an interrupted blocked attempt, not an automatic wake, autonomous responder, uninterrupted run, completed diagnostic checks, review or product delivery, or OS isolation. The two settled requests are not to be re-disposed. The source guidance followed this event and did not cause it. Source checks and independent review are recorded in the delivery PR |
 | Lead-prepared inputs (0.1.16 candidate) | 2026-10-15: separately, an OPD worker's archive preparation waited 15 m 12 s for approval and was resolved once in the native UI (operator report, not reproduced here). In this [Factory #32](https://github.com/arcitai/factory/issues/32) trial the lead prepared harmless inputs in the worker's task worktree; the same Claude worker used native Read and Edit for one result-file change and the AgentOps text change. The lead independently verified the inputs unchanged and the expected result hash; the run ended terminal with nothing pending, and its native timeline showed no approval request or worker shell command | One same-worker prepared-input text task, not a fresh-context comparison, measured wait reduction, general prompt-free operation, worker build/test permission, permission wake or decision authority, or isolation. A separate compound Python edit and Node/shell check request in that same OPD run remains held at this observation; its lead has a recovery handoff, not a passing recovery. Source checks and independent review are recorded in the delivery PR |
 | Foundation readiness handoff (0.1.12 candidate) | 2026-10-11: [Factory #27](https://github.com/arcitai/factory/issues/27) source change after an AIOS assessment accepted a stale mandatory authoring self-dependency, its disabled private reporting was found only by readback, its lead could not read its private record, and Project1's current README kept stale pre-delivery claims (coordinator observations, not reproduced here). Package, unit and staging checks only | Behavioral evidence and independent review are pending in the delivery PR. Native binding/authentication behavior stays with [Factory #3](https://github.com/arcitai/factory/issues/3); Markdown does not repair it |