execweave 0.7.4__tar.gz → 0.7.5__tar.gz

execweave-0.7.5/.github/workflows/provider-capability-stage-integrity.yml

@@ -0,0 +1,42 @@
+name: Provider Capability Stage Integrity
+
+# No path filter. This job guards release red lines — the disabled CONNECT handler,
+# the existing test set, the conversation completeness vocabulary, release metadata —
+# so it must run on every pull request. A filter listing only the capability-probe
+# files would arm the guard exactly when the guarded files are NOT the ones changing:
+# a pull request that weakens http_proxy.py, deletes a test, or bumps the version
+# while touching none of those paths would skip this job entirely.
+on:
+  pull_request:
+  workflow_dispatch:
+    inputs:
+      baseline_ref:
+        description: Base commit or ref used to reconstruct baseline test node IDs
+        required: true
+        type: string
+
+permissions:
+  contents: read
+
+jobs:
+  integrity:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Check out pull request
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+          cache: pip
+
+      - name: Install
+        run: python -m pip install -e ".[dev]"
+
+      - name: Validate stage integrity
+        run: >-
+          python scripts/check_release_stage_integrity.py
+          --baseline-ref "${{ github.event.pull_request.base.sha || inputs.baseline_ref }}"

{execweave-0.7.4 → execweave-0.7.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: execweave
-Version: 0.7.4
+Version: 0.7.5
 Summary: Turn AI-agent runtime behavior into a local interactive execution graph.
 Project-URL: Homepage, https://github.com/Irish-kw/ExecWeave
 Project-URL: Repository, https://github.com/Irish-kw/ExecWeave
@@ -82,7 +82,7 @@ Install the latest published wheel/sdist from PyPI:
 python -m pip install -U execweave
 ```
 
-The current release is **v0.7.4**.
+The current release is **v0.7.5**.
 
 For development:
 
@@ -118,7 +118,7 @@ execweave record --open -- python my_agent.py
 
 `execweave top -- codex` keeps the Agent interactive in the launch terminal while opening/attaching the detached Top dashboard according to the host environment.
 
-**v0.7.4 — per-agent conversation focus in the dashboard.** Building on the provider-neutral, agent-local multi-agent conversations materialized in v0.7.3, selecting an agent node now scopes the conversation panel to that agent alone, with its visible item count and a control to restore the full tree. A click resolves to a thread by graph agent node identity rather than by label matching, so agents that share a provider nickname stay distinct. Markdown conversation sections now lead with the agent path instead of the provider nickname, so every section names the agent it belongs to. Both the standalone viewer and the live dashboard are covered; conversation materialization itself is unchanged.
+**v0.7.5 — per-agent conversation isolation by default, and an evidence availability matrix.** Building on the provider-neutral, agent-local multi-agent conversations materialized in earlier releases, opening a run now shows one self-contained section per agent: each holds the turns that agent sent and the turns addressed to it, so a child keeps its own assignment and answer while a sibling's exchange never appears. Field-level availability is recorded separately from conversation completeness, so an encrypted reasoning field no longer drags an otherwise archived transcript down. An offline capability probe reports, per client and surface, what each provider actually exposes, and records what cannot be read without claiming to know why.
 
 The unified dashboard brings the execution graph, logs, and conversation records into the same inspection flow. Finalized runs generate `conversations.md` and `conversations.json`, while validated provider transcripts are copied into the run-local SHA-256 content store. Claude Code, OpenAI Codex, Cursor, OpenCode, and Google Antigravity use the strongest provider-exposed multi-agent evidence available to each integration. For gateways and local runtimes that expose only root request/response traffic, ExecWeave shows only that root conversation and does not invent subagents or hidden routing.
 
@@ -327,7 +327,7 @@ Conversation isolation is an attribution/display rule, not a redaction boundary.
 
 ## Current status
 
-v0.7.4 combines cross-platform runtime collection, materialized execution graphs, standalone/live dashboards, conservative provider↔runtime correlation, full-fidelity content-addressed provider evidence, attributable multi-agent execution traces, direct run-local conversation access, agent-local conversation isolation across provider-neutral projections, and per-agent conversation focus in the standalone and live dashboards. Supported integrations preserve the strongest identity/routing evidence actually exposed by each provider and abstain when that evidence is unavailable. Observed evidence and inference remain separate by design.
+v0.7.5 combines cross-platform runtime collection, materialized execution graphs, standalone/live dashboards, conservative provider↔runtime correlation, full-fidelity content-addressed provider evidence, attributable multi-agent execution traces, direct run-local conversation access, agent-local conversation isolation across provider-neutral projections, and per-agent conversation focus in the standalone and live dashboards. Supported integrations preserve the strongest identity/routing evidence actually exposed by each provider and abstain when that evidence is unavailable. Observed evidence and inference remain separate by design.
 
 ## Documentation
 

{execweave-0.7.4 → execweave-0.7.5}/README.de.md

@@ -31,7 +31,7 @@ Installieren Sie das neueste veröffentlichte wheel/sdist von PyPI:
 python -m pip install -U execweave
 ```
 
-Die aktuelle Version ist **v0.7.4**.
+Die aktuelle Version ist **v0.7.5**.
 
 Für die Entwicklung:
 
@@ -67,7 +67,7 @@ execweave record --open -- python my_agent.py
 
 `execweave top -- codex` hält den Agent im Startterminal interaktiv und öffnet bzw. verbindet gleichzeitig das detached Top dashboard entsprechend der Host-Umgebung.
 
-**v0.7.4 — Conversation-Fokus pro Agent im Dashboard.** Aufbauend auf den in v0.7.3 materialisierten provider-neutralen, agent-lokalen Multi-Agent-Conversations begrenzt die Auswahl eines agent node das Conversation-Panel jetzt auf genau diesen Agent, samt Anzahl der sichtbaren Einträge und einem Control zum Wiederherstellen des vollständigen Baums. Ein Klick wird über die identity des graph agent node und nicht über Label-Vergleich zu einem Thread aufgelöst, sodass Agents mit demselben Provider-Nickname unterscheidbar bleiben. Markdown-Conversation-Abschnitte beginnen nun mit dem agent path statt mit dem Provider-Nickname, sodass jeder Abschnitt den zugehörigen Agent benennt. Sowohl der Standalone-Viewer als auch das Live-Dashboard sind abgedeckt; die Conversation-Materialisierung selbst bleibt unverändert.
+**v0.7.5 — Conversation-Isolation pro Agent als Standard und eine Evidence-Availability-Matrix.** Beim Öffnen eines Runs hat jeder Agent nun einen eigenen Abschnitt: Er enthält die Turns, die dieser Agent gesendet hat, und die an ihn gerichteten, sodass ein Child seine eigene Assignment und Antwort behält, während ein Austausch zwischen Siblings nie erscheint. Field-level availability wird getrennt von der conversation completeness erfasst, sodass ein verschlüsseltes Reasoning-Feld ein ansonsten vollständig archiviertes Transcript nicht mehr herabstuft. Eine Offline-Capability-Probe meldet je Client und Surface, was der Provider tatsächlich offenlegt, und hält Nichtlesbares fest, ohne den Grund zu behaupten.
 
 Das vereinheitlichte Dashboard bringt execution graph, logs und conversation records in denselben Inspection-Flow. Finalisierte Runs erzeugen `conversations.md` und `conversations.json`; validierte Provider-Transcripts werden in den run-local SHA-256 content store kopiert. Claude Code, OpenAI Codex, Cursor, OpenCode und Google Antigravity verwenden jeweils die stärkste Multi-Agent-Evidence, die ihre Integration tatsächlich offenlegt. Wenn ein Gateway oder lokaler Runtime nur root request/response sichtbar macht, zeigt ExecWeave nur diese Root-Conversation und erfindet weder Subagents noch hidden routing.
 
@@ -276,7 +276,7 @@ Conversation isolation ist eine Attribution-/Display-Regel, keine Redaction Boun
 
 ## Aktueller Stand
 
-v0.7.4 kombiniert cross-platform runtime collection, materialized execution graphs, standalone/live dashboards, konservative provider↔runtime correlation, full-fidelity content-addressed provider evidence, attributable multi-agent execution traces, direkten run-local conversation access agent-local conversation isolation in provider-neutralen Projektionen und per-agent conversation focus in den Standalone- und Live-Dashboards. Jede Integration bewahrt nur die stärkste identity/routing evidence, die der Provider tatsächlich offenlegt, und enthält sich, wenn diese Evidence fehlt. Observed evidence und inference bleiben konstruktiv getrennt.
+v0.7.5 kombiniert cross-platform runtime collection, materialized execution graphs, standalone/live dashboards, konservative provider↔runtime correlation, full-fidelity content-addressed provider evidence, attributable multi-agent execution traces, direkten run-local conversation access agent-local conversation isolation in provider-neutralen Projektionen und per-agent conversation focus in den Standalone- und Live-Dashboards. Jede Integration bewahrt nur die stärkste identity/routing evidence, die der Provider tatsächlich offenlegt, und enthält sich, wenn diese Evidence fehlt. Observed evidence und inference bleiben konstruktiv getrennt.
 
 ## Dokumentation
 

{execweave-0.7.4 → execweave-0.7.5}/README.fr.md

@@ -31,7 +31,7 @@ Installez le dernier wheel/sdist publié depuis PyPI :
 python -m pip install -U execweave
 ```
 
-La version courante est **v0.7.4**.
+La version courante est **v0.7.5**.
 
 Pour le développement :
 
@@ -67,7 +67,7 @@ execweave record --open -- python my_agent.py
 
 `execweave top -- codex` garde l’Agent interactif dans le terminal de lancement tout en ouvrant ou en attachant le dashboard Top détaché selon l’environnement hôte.
 
-**v0.7.4 — focus de conversation par agent dans le dashboard.** En s’appuyant sur les conversations multi-agent provider-neutral et agent-local matérialisées en v0.7.3, sélectionner un agent node restreint désormais le panneau de conversation à cet agent seul, avec le nombre d’éléments visibles et un contrôle pour restaurer l’arbre complet. Un clic est résolu en thread via l’identity du graph agent node et non par correspondance de label, de sorte que des agents partageant un même provider nickname restent distincts. Les sections de conversation Markdown commencent maintenant par l’agent path plutôt que par le provider nickname, si bien que chaque section nomme l’agent auquel elle appartient. Le viewer standalone et le live dashboard sont tous deux couverts ; la matérialisation des conversations est inchangée.
+**v0.7.5 — isolation des conversations par agent par défaut, et matrice de disponibilité des evidence.** À l’ouverture d’un run, chaque agent dispose désormais de sa propre section : elle contient les turns que cet agent a envoyés et ceux qui lui étaient adressés, si bien qu’un child conserve son assignment et sa réponse tandis qu’un échange entre siblings n’apparaît jamais. La field-level availability est enregistrée séparément de la conversation completeness, de sorte qu’un champ reasoning chiffré ne dégrade plus un transcript par ailleurs archivé. Une capability probe hors ligne rapporte, par client et par surface, ce que chaque provider expose réellement, et consigne ce qui reste illisible sans prétendre en connaître la raison.
 
 Le dashboard unifié rassemble execution graph, logs et conversation records dans le même flux d’inspection. Les runs finalisés génèrent `conversations.md` et `conversations.json`, et les transcripts provider validés sont copiés dans le content store SHA-256 local au run. Claude Code, OpenAI Codex, Cursor, OpenCode et Google Antigravity exploitent chacun le niveau de multi-agent evidence réellement exposé par leur integration. Lorsqu’un gateway ou un local runtime n’expose que des request/response root, ExecWeave affiche uniquement cette root conversation et n’invente ni subagent ni hidden routing.
 
@@ -276,7 +276,7 @@ La conversation isolation est une règle d’attribution et d’affichage, pas u
 
 ## État actuel
 
-v0.7.4 combine cross-platform runtime collection, materialized execution graphs, standalone/live dashboards, conservative provider↔runtime correlation, full-fidelity content-addressed provider evidence, attributable multi-agent execution traces, accès direct aux conversations run-local agent-local conversation isolation dans les projections provider-neutral et per-agent conversation focus dans les dashboards standalone et live. Chaque integration conserve uniquement la meilleure identity/routing evidence réellement exposée par le provider et s’abstient lorsque cette evidence manque. Observed evidence et inference restent séparées par conception.
+v0.7.5 combine cross-platform runtime collection, materialized execution graphs, standalone/live dashboards, conservative provider↔runtime correlation, full-fidelity content-addressed provider evidence, attributable multi-agent execution traces, accès direct aux conversations run-local agent-local conversation isolation dans les projections provider-neutral et per-agent conversation focus dans les dashboards standalone et live. Chaque integration conserve uniquement la meilleure identity/routing evidence réellement exposée par le provider et s’abstient lorsque cette evidence manque. Observed evidence et inference restent séparées par conception.
 
 ## Documentation
 

{execweave-0.7.4 → execweave-0.7.5}/README.ja.md

@@ -31,7 +31,7 @@ PyPI から最新の wheel/sdist をインストールします。
 python -m pip install -U execweave
 ```
 
-現在のリリースは **v0.7.4** です。
+現在のリリースは **v0.7.5** です。
 
 開発環境では次のようにインストールできます。
 
@@ -67,7 +67,7 @@ execweave record --open -- python my_agent.py
 
 `execweave top -- codex` は Agent を起動 terminal 上でインタラクティブなまま保ち、ホスト環境に応じて detached Top dashboard を開くか既存のものに attach します。
 
-**v0.7.4 — dashboard での per-agent conversation focus。** v0.7.3 で materialize された provider-neutral かつ agent-local な multi-agent conversation を土台に、agent node を選択すると conversation panel がその agent だけに絞り込まれ、表示項目数と全体ツリーに戻すコントロールが付きます。クリックは label の一致ではなく graph agent node identity で thread に解決されるため、同じ provider nickname を共有する agent どうしも混ざりません。Markdown の conversation セクションは provider nickname ではなく agent path で始まるようになり、各セクションがどの agent のものかを明示します。standalone viewer と live dashboard の両方が対象で、conversation materialization 自体は変更していません。
+**v0.7.5 — 既定で per-agent conversation isolation、そして evidence availability matrix。** run を開くと各 agent が独立したセクションを持ち、その agent が送った turn と宛てられた turn だけを含みます。child は自身の assignment と answer を保持し、sibling 同士のやり取りは現れません。field-level availability は conversation completeness と分けて記録されるため、reasoning フィールドが暗号化されていても、完全に保存された transcript が一緒に降格されることはありません。オフラインの capability probe は client と surface ごとに provider が実際に公開したものを報告し、読み取れないものは理由を断定せずに記録します。
 
 統合 dashboard では execution graph、logs、conversation records を同じ inspection flow で確認できます。Finalized run は `conversations.md` と `conversations.json` を生成し、検証済み provider transcript は run-local SHA-256 content store にコピーされます。Claude Code、OpenAI Codex、Cursor、OpenCode、Google Antigravity は、それぞれが実際に公開する最も強い multi-agent evidence を利用します。gateway や local runtime が root request/response しか公開しない場合、ExecWeave は root conversation だけを表示し、subagent や hidden routing を捏造しません。
 
@@ -276,7 +276,7 @@ Conversation isolation は attribution/display の規則であり redaction boun
 
 ## 現在の状態
 
-v0.7.4 は cross-platform runtime collection、materialized execution graph、standalone/live dashboard、保守的な provider↔runtime correlation、content-addressed full-fidelity provider evidence、attributable multi-agent execution trace、run-local conversation access、provider-neutral projection 上の agent-local conversation isolation、standalone および live dashboard での per-agent conversation focus を統合します。各 integration は provider が実際に公開した最も強い identity/routing evidence のみを保持し、不足する場合は abstain します。Observed evidence と inference は設計上分離されたままです。
+v0.7.5 は cross-platform runtime collection、materialized execution graph、standalone/live dashboard、保守的な provider↔runtime correlation、content-addressed full-fidelity provider evidence、attributable multi-agent execution trace、run-local conversation access、provider-neutral projection 上の agent-local conversation isolation、standalone および live dashboard での per-agent conversation focus を統合します。各 integration は provider が実際に公開した最も強い identity/routing evidence のみを保持し、不足する場合は abstain します。Observed evidence と inference は設計上分離されたままです。
 
 ## ドキュメント
 

{execweave-0.7.4 → execweave-0.7.5}/README.ko.md

@@ -31,7 +31,7 @@ PyPI에서 최신 wheel/sdist를 설치합니다.
 python -m pip install -U execweave
 ```
 
-현재 릴리스는 **v0.7.4** 입니다.
+현재 릴리스는 **v0.7.5** 입니다.
 
 개발 설치:
 
@@ -67,7 +67,7 @@ execweave record --open -- python my_agent.py
 
 `execweave top -- codex`는 Agent를 시작 terminal에서 interactive 상태로 유지하면서 host 환경에 따라 detached Top dashboard를 열거나 attach합니다.
 
-**v0.7.4 — dashboard의 per-agent conversation focus.** v0.7.3에서 materialize된 provider-neutral, agent-local multi-agent conversation을 토대로, 이제 agent node를 선택하면 conversation panel이 해당 agent만 보여주며 표시 항목 수와 전체 트리로 복원하는 컨트롤이 함께 제공됩니다. 클릭은 label 비교가 아니라 graph agent node identity로 thread에 매핑되므로 동일한 provider nickname을 공유하는 agent끼리도 섞이지 않습니다. Markdown conversation 섹션은 provider nickname 대신 agent path로 시작하여 각 섹션이 어떤 agent의 것인지 명시합니다. standalone viewer와 live dashboard 모두 적용되며 conversation materialization 자체는 변경되지 않았습니다.
+**v0.7.5 — 기본값이 된 per-agent conversation isolation, 그리고 evidence availability matrix.** run을 열면 각 agent가 독립된 섹션을 가지며, 그 agent가 보낸 turn과 그 agent에게 전달된 turn만 포함합니다. child는 자신의 assignment와 answer를 유지하고 sibling 간의 교환은 나타나지 않습니다. field-level availability는 conversation completeness와 분리되어 기록되므로 reasoning 필드가 암호화되어도 완전히 보관된 transcript가 함께 강등되지 않습니다. 오프라인 capability probe는 client와 surface별로 provider가 실제 노출한 것을 보고하고, 읽을 수 없는 것은 이유를 단정하지 않고 기록합니다.
 
 통합 dashboard는 execution graph, logs, conversation records를 하나의 inspection flow에서 제공합니다. Finalized run은 `conversations.md`와 `conversations.json`을 만들며, 검증된 provider transcript는 run-local SHA-256 content store로 복사됩니다. Claude Code, OpenAI Codex, Cursor, OpenCode, Google Antigravity는 각 integration이 실제로 노출하는 가장 강한 multi-agent evidence를 사용합니다. gateway나 local runtime이 root request/response만 노출하면 ExecWeave는 root conversation만 보여 주며 subagent나 hidden routing을 만들어내지 않습니다.
 
@@ -276,7 +276,7 @@ Conversation isolation은 attribution/display 규칙이지 redaction boundary가
 
 ## 현재 상태
 
-v0.7.4는 cross-platform runtime collection, materialized execution graph, standalone/live dashboard, 보수적인 provider↔runtime correlation, content-addressed full-fidelity provider evidence, attributable multi-agent execution trace, run-local conversation access, provider-neutral projection의 agent-local conversation isolation, standalone 및 live dashboard의 per-agent conversation focus를 통합합니다. 각 integration은 provider가 실제로 노출한 가장 강한 identity/routing evidence만 보존하고 충분한 증거가 없으면 abstain합니다. Observed evidence와 inference는 설계상 계속 분리됩니다.
+v0.7.5는 cross-platform runtime collection, materialized execution graph, standalone/live dashboard, 보수적인 provider↔runtime correlation, content-addressed full-fidelity provider evidence, attributable multi-agent execution trace, run-local conversation access, provider-neutral projection의 agent-local conversation isolation, standalone 및 live dashboard의 per-agent conversation focus를 통합합니다. 각 integration은 provider가 실제로 노출한 가장 강한 identity/routing evidence만 보존하고 충분한 증거가 없으면 abstain합니다. Observed evidence와 inference는 설계상 계속 분리됩니다.
 
 ## 문서
 

{execweave-0.7.4 → execweave-0.7.5}/README.md

@@ -31,7 +31,7 @@ Install the latest published wheel/sdist from PyPI:
 python -m pip install -U execweave
 ```
 
-The current release is **v0.7.4**.
+The current release is **v0.7.5**.
 
 For development:
 
@@ -67,7 +67,7 @@ execweave record --open -- python my_agent.py
 
 `execweave top -- codex` keeps the Agent interactive in the launch terminal while opening/attaching the detached Top dashboard according to the host environment.
 
-**v0.7.4 — per-agent conversation focus in the dashboard.** Building on the provider-neutral, agent-local multi-agent conversations materialized in v0.7.3, selecting an agent node now scopes the conversation panel to that agent alone, with its visible item count and a control to restore the full tree. A click resolves to a thread by graph agent node identity rather than by label matching, so agents that share a provider nickname stay distinct. Markdown conversation sections now lead with the agent path instead of the provider nickname, so every section names the agent it belongs to. Both the standalone viewer and the live dashboard are covered; conversation materialization itself is unchanged.
+**v0.7.5 — per-agent conversation isolation by default, and an evidence availability matrix.** Building on the provider-neutral, agent-local multi-agent conversations materialized in earlier releases, opening a run now shows one self-contained section per agent: each holds the turns that agent sent and the turns addressed to it, so a child keeps its own assignment and answer while a sibling's exchange never appears. Field-level availability is recorded separately from conversation completeness, so an encrypted reasoning field no longer drags an otherwise archived transcript down. An offline capability probe reports, per client and surface, what each provider actually exposes, and records what cannot be read without claiming to know why.
 
 The unified dashboard brings the execution graph, logs, and conversation records into the same inspection flow. Finalized runs generate `conversations.md` and `conversations.json`, while validated provider transcripts are copied into the run-local SHA-256 content store. Claude Code, OpenAI Codex, Cursor, OpenCode, and Google Antigravity use the strongest provider-exposed multi-agent evidence available to each integration. For gateways and local runtimes that expose only root request/response traffic, ExecWeave shows only that root conversation and does not invent subagents or hidden routing.
 
@@ -276,7 +276,7 @@ Conversation isolation is an attribution/display rule, not a redaction boundary.
 
 ## Current status
 
-v0.7.4 combines cross-platform runtime collection, materialized execution graphs, standalone/live dashboards, conservative provider↔runtime correlation, full-fidelity content-addressed provider evidence, attributable multi-agent execution traces, direct run-local conversation access, agent-local conversation isolation across provider-neutral projections, and per-agent conversation focus in the standalone and live dashboards. Supported integrations preserve the strongest identity/routing evidence actually exposed by each provider and abstain when that evidence is unavailable. Observed evidence and inference remain separate by design.
+v0.7.5 combines cross-platform runtime collection, materialized execution graphs, standalone/live dashboards, conservative provider↔runtime correlation, full-fidelity content-addressed provider evidence, attributable multi-agent execution traces, direct run-local conversation access, agent-local conversation isolation across provider-neutral projections, and per-agent conversation focus in the standalone and live dashboards. Supported integrations preserve the strongest identity/routing evidence actually exposed by each provider and abstain when that evidence is unavailable. Observed evidence and inference remain separate by design.
 
 ## Documentation
 

{execweave-0.7.4 → execweave-0.7.5}/README.ru.md

@@ -31,7 +31,7 @@ ExecWeave — source-available, local-first проект наблюдаемос
 python -m pip install -U execweave
 ```
 
-Текущая версия — **v0.7.4**.
+Текущая версия — **v0.7.5**.
 
 Для разработки:
 
@@ -67,7 +67,7 @@ execweave record --open -- python my_agent.py
 
 `execweave top -- codex` оставляет Agent интерактивным в стартовом terminal и одновременно открывает либо подключает detached Top dashboard в зависимости от host environment.
 
-**v0.7.4 — фокус на conversation отдельного agent в dashboard.** Опираясь на provider-neutral, agent-local multi-agent conversations, материализованные в v0.7.3, выбор agent node теперь ограничивает conversation panel только этим agent, показывая число видимых элементов и control для возврата к полному дереву. Клик сопоставляется с thread по identity graph agent node, а не по совпадению label, поэтому agents с одинаковым provider nickname остаются различимыми. Секции conversation в Markdown теперь начинаются с agent path, а не с provider nickname, так что каждая секция называет свой agent. Покрыты и standalone viewer, и live dashboard; сама материализация conversation не изменена.
+**v0.7.5 — изоляция conversation по agent по умолчанию и матрица доступности evidence.** При открытии run каждый agent теперь имеет собственную секцию: в неё входят turns, отправленные этим agent, и те, что были адресованы ему, поэтому child сохраняет собственные assignment и ответ, а обмен между siblings не появляется никогда. Field-level availability фиксируется отдельно от conversation completeness, поэтому зашифрованное поле reasoning больше не понижает полностью архивированный transcript. Офлайновый capability probe сообщает по каждому client и surface, что provider действительно раскрывает, и фиксирует нечитаемое, не утверждая причину.
 
 Единый dashboard объединяет execution graph, logs и conversation records в одном inspection flow. Finalized runs создают `conversations.md` и `conversations.json`, а проверенные provider transcripts копируются в run-local SHA-256 content store. Claude Code, OpenAI Codex, Cursor, OpenCode и Google Antigravity используют наиболее сильную multi-agent evidence, которую реально раскрывает соответствующая integration. Если gateway или local runtime показывает только root request/response, ExecWeave отображает только root conversation и не выдумывает subagents или hidden routing.
 
@@ -276,7 +276,7 @@ Conversation isolation — это правило attribution/display, а не re
 
 ## Текущий статус
 
-v0.7.4 объединяет cross-platform runtime collection, materialized execution graphs, standalone/live dashboards, conservative provider↔runtime correlation, full-fidelity content-addressed provider evidence, attributable multi-agent execution traces, прямой run-local conversation access agent-local conversation isolation в provider-neutral projections и per-agent conversation focus в standalone и live dashboards. Каждая integration сохраняет только наиболее сильную identity/routing evidence, реально раскрытую provider, и abstain-ит при её отсутствии. Observed evidence и inference по-прежнему разделены архитектурно.
+v0.7.5 объединяет cross-platform runtime collection, materialized execution graphs, standalone/live dashboards, conservative provider↔runtime correlation, full-fidelity content-addressed provider evidence, attributable multi-agent execution traces, прямой run-local conversation access agent-local conversation isolation в provider-neutral projections и per-agent conversation focus в standalone и live dashboards. Каждая integration сохраняет только наиболее сильную identity/routing evidence, реально раскрытую provider, и abstain-ит при её отсутствии. Observed evidence и inference по-прежнему разделены архитектурно.
 
 ## Документация
 

{execweave-0.7.4 → execweave-0.7.5}/README.zh-CN.md

@@ -31,7 +31,7 @@ ExecWeave 是一个 source-available、local-first 的可观测性项目，把 A
 python -m pip install -U execweave
 ```
 
-当前正式版本是 **v0.7.4**。
+当前正式版本是 **v0.7.5**。
 
 开发安装：
 
@@ -67,7 +67,7 @@ execweave record --open -- python my_agent.py
 
 `execweave top -- codex` 会让 Agent 保持在启动 terminal 中交互，并根据主机环境打开或附加 detached Top dashboard。
 
-**v0.7.4 — dashboard 的 per-agent conversation focus。** 延续 v0.7.3 已经 materialize 的 provider-neutral、agent-local multi-agent conversation，现在点选某个 agent node，conversation panel 只会显示该 agent 自己的 thread，并附带可见项目数与还原完整树状视图的控件。点击是以 graph agent node identity 对应到 thread，而不是比对 label，因此即使多个 agent 共用同一个 provider nickname 也不会被混在一起。Markdown 的 conversation 章节现在以 agent path 开头，而不是 provider nickname，每个章节都能明确指出自己属于哪一个 agent。standalone viewer 与 live dashboard 都覆盖；conversation materialization 本身没有变动。
+**v0.7.5 — 默认的 per-agent conversation isolation，以及 evidence availability matrix。** 打开一次 run，现在每个 agent 各有一个独立区块：只包含该 agent 发出的与寄给它的 turn，因此 child 保留自己的 assignment 与 answer，而 sibling 之间的往来完全不会出现。field-level availability 与 conversation completeness 分开记录，因此某个 reasoning 字段被加密，不会把已完整归档的 transcript 一起降级。离线的 capability probe 会按 client 与 surface 报告各 provider 实际暴露了什么，并记录读不到的部分，而不宣称知道原因。
 
 统一 dashboard 把 execution graph、logs 与 conversation records 放进同一条 inspection flow。Finalized run 会生成 `conversations.md` 与 `conversations.json`，经过验证的 provider transcript 也会复制进 run-local SHA-256 content store。Claude Code、OpenAI Codex、Cursor、OpenCode 与 Google Antigravity 都依据各自实际暴露的 evidence 强度建立 multi-agent trace；如果 gateway 或 local runtime 只提供 root request/response，ExecWeave 就只显示 root conversation，不会虚构 subagent 或 hidden routing。
 
@@ -276,7 +276,7 @@ Conversation isolation 是 attribution/display 规则，而不是 redaction boun
 
 ## 当前状态
 
-v0.7.4 组合 cross-platform runtime collection、materialized execution graph、standalone/live dashboard、保守的 provider↔runtime correlation、content-addressed full-fidelity provider evidence、可归属的 multi-agent execution trace、run-local conversation access，provider-neutral projection 上的 agent-local conversation isolation，以及 standalone 与 live dashboard 上的 per-agent conversation focus。各 integration 只保留 provider 实际暴露的最强 identity/routing evidence，证据不足时选择 abstain。Observed evidence 与 inference 从设计上保持分离。
+v0.7.5 组合 cross-platform runtime collection、materialized execution graph、standalone/live dashboard、保守的 provider↔runtime correlation、content-addressed full-fidelity provider evidence、可归属的 multi-agent execution trace、run-local conversation access，provider-neutral projection 上的 agent-local conversation isolation，以及 standalone 与 live dashboard 上的 per-agent conversation focus。各 integration 只保留 provider 实际暴露的最强 identity/routing evidence，证据不足时选择 abstain。Observed evidence 与 inference 从设计上保持分离。
 
 ## 文档
 

{execweave-0.7.4 → execweave-0.7.5}/README.zh-TW.md

@@ -31,7 +31,7 @@ ExecWeave 是一個 source-available、local-first 的可觀測性專案，會
 python -m pip install -U execweave
 ```
 
-目前正式版本是 **v0.7.4**。
+目前正式版本是 **v0.7.5**。
 
 開發安裝：
 
@@ -67,7 +67,7 @@ execweave record --open -- python my_agent.py
 
 `execweave top -- codex` 會讓 Agent 保持在啟動 terminal 中互動，並依主機環境開啟或附加 detached Top dashboard。
 
-**v0.7.4 — dashboard 的 per-agent conversation focus。** 延續 v0.7.3 已經 materialize 的 provider-neutral、agent-local multi-agent conversation，現在點選某個 agent node，conversation panel 會只顯示該 agent 自己的 thread，並附上可見項目數與還原完整樹狀檢視的控制項。點擊是以 graph agent node identity 對應到 thread，而不是比對 label，所以即使多個 agent 共用同一個 provider nickname 也不會被混在一起。Markdown 的 conversation 章節現在以 agent path 開頭，而不是 provider nickname，每個章節都能明確指出自己屬於哪一個 agent。standalone viewer 與 live dashboard 都涵蓋；conversation materialization 本身沒有變動。
+**v0.7.5 — 預設的 per-agent conversation isolation，以及 evidence availability matrix。** 打開一次 run，現在每個 agent 各有一個獨立區塊：只包含該 agent 送出的與寄給它的 turn，所以 child 保留自己的 assignment 與 answer，而 sibling 之間的往來完全不會出現。field-level availability 與 conversation completeness 分開記錄，因此某個 reasoning 欄位被加密，不會把已完整封存的 transcript 一起降級。離線的 capability probe 會依 client 與 surface 回報各 provider 實際曝露了什麼，並記錄讀不到的部分，而不宣稱知道原因。
 
 統一 dashboard 把 execution graph、logs 與 conversation records 放進同一條 inspection flow。Finalized run 會產生 `conversations.md` 與 `conversations.json`，經驗證的 provider transcript 也會複製進 run-local SHA-256 content store。Claude Code、OpenAI Codex、Cursor、OpenCode 與 Google Antigravity 都依各自實際曝露的 evidence 強度建立 multi-agent trace；若 gateway 或 local runtime 只提供 root request/response，ExecWeave 就只顯示 root conversation，不會虛構 subagent 或 hidden routing。
 
@@ -276,7 +276,7 @@ Conversation isolation 是 attribution/display 規則，不是 redaction boundar
 
 ## 目前狀態
 
-v0.7.4 整合 cross-platform runtime collection、materialized execution graph、standalone/live dashboard、保守的 provider↔runtime correlation、content-addressed full-fidelity provider evidence、可歸屬的 multi-agent execution trace、run-local conversation access，provider-neutral projection 上的 agent-local conversation isolation，以及 standalone 與 live dashboard 上的 per-agent conversation focus。各 integration 只保留 provider 實際曝露的最強 identity/routing evidence，證據不足時選擇 abstain。Observed evidence 與 inference 仍從設計上分離。
+v0.7.5 整合 cross-platform runtime collection、materialized execution graph、standalone/live dashboard、保守的 provider↔runtime correlation、content-addressed full-fidelity provider evidence、可歸屬的 multi-agent execution trace、run-local conversation access，provider-neutral projection 上的 agent-local conversation isolation，以及 standalone 與 live dashboard 上的 per-agent conversation focus。各 integration 只保留 provider 實際曝露的最強 identity/routing evidence，證據不足時選擇 abstain。Observed evidence 與 inference 仍從設計上分離。
 
 ## 文件
 

execweave-0.7.5/docs/internal/provider-capability-matrix.md

@@ -0,0 +1,101 @@
+# Provider Capability Matrix (0.7.5)
+
+This document records the evidence contract for the 0.7.5 capability probe. It is an
+internal implementation note, not a release-support claim.
+
+## Scope
+
+The probe is offline. It reads only caller-supplied existing JSON/JSONL artifacts and
+repository source used for the narrow Codex local-decryptor marker audit. It performs no
+network capture, endpoint rewriting, TLS interception, decryption, or provider login.
+
+The machine-readable inventory and matrix are emitted by:
+
+```bash
+python scripts/probe_provider_capability.py --inventory-only
+python scripts/probe_provider_capability.py
+```
+
+A concrete artifact is attached to exactly one inventory combination:
+
+```bash
+python scripts/probe_provider_capability.py \
+  --artifact codex-cli:subscription=tests/fixtures/codex_multi_agent/rollout-main.jsonl
+```
+
+When a client has multiple auth modes or surfaces, the selector must name the ambiguous
+part explicitly. A missing artifact never deletes the row: every required field remains
+in the matrix as `not_observed` with a reason.
+
+## Required inventory
+
+Tier A is release-blocking and contains these clients:
+
+- Codex CLI
+- Claude Code
+- Gemini CLI
+- Cursor Agent
+- OpenCode
+- Ollama
+
+Every combination defined by the inventory probes at least: `system`, `prompt`,
+`messages`, `tool_definitions`, `tool_arguments`, `tool_results`, `assistant_output`,
+`reasoning`, and `usage`.
+
+Tier B is environment-dependent and currently covers Cursor autocomplete/Tab plus local
+LM Studio, llama.cpp, and vLLM surfaces. Untested Tier B rows remain explicit rather than
+being silently omitted.
+
+## Evidence boundaries
+
+Conversation completeness remains owned by `agent_topology.py` and is not changed by
+this stage. Field-level availability is defined separately in
+`evidence_availability.py`. Therefore this is valid:
+
+```text
+conversation_completeness = provider_transcript
+reasoning_availability    = opaque_encrypted
+```
+
+A single unavailable or encrypted field must never downgrade an otherwise archived
+provider transcript.
+
+`available` and `complete_from_surface` have fixed precedence:
+
+- `available`: content is readable, but the probe cannot prove the full surface value was
+  preserved.
+- `complete_from_surface`: the surface observation boundary is known and the complete
+  value supplied by that surface was preserved.
+- If both appear applicable, `complete_from_surface` wins. They are never emitted
+  simultaneously for one field observation.
+
+`complete_from_surface` means complete from the observed integration surface. It does
+not claim visibility into hidden model/provider state.
+
+## Codex encrypted reasoning
+
+The checked-in Codex rollout fixture contains a reasoning item with
+`encrypted_content` beginning with the Fernet-shaped `gAAAAA` prefix. That is direct
+evidence only that an opaque encrypted payload was observed.
+
+The repository source audit looks for a small set of Codex-local decryptor markers. If
+none are observed, the matrix may state:
+
+```text
+availability   = opaque_encrypted
+decryptability = no_local_decryptor_observed
+```
+
+This is an absence observation. It must never be upgraded to `server-side-only`, or to
+`provider_documented_unavailable`, without provider documentation or direct verifiable
+evidence.
+
+## Output semantics
+
+Each matrix row records client, tested client version when directly present in the
+artifact, provider, auth mode, surface, transport mode, field, availability,
+decryptability, evidence source, evidence strength, tier, and notes.
+
+The probe stores paths and classifications, not copied prompt/tool/reasoning content.
+This stage is capability discovery only; it does not add a second full-fidelity storage
+or conversation-ingestion path.

{execweave-0.7.4 → execweave-0.7.5}/pyproject.toml

@@ -4,8 +4,8 @@ build-backend = "hatchling.build"
 
 [project]
 name = "execweave"
-# ExecWeave v0.7.4 release metadata
-version = "0.7.4"
+# ExecWeave v0.7.5 release metadata
+version = "0.7.5"
 description = "Turn AI-agent runtime behavior into a local interactive execution graph."
 readme = "README.md"
 requires-python = ">=3.10"