@oneciel-ai/claude-any 0.1.29 → 0.1.30

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -44,7 +44,7 @@ arguments through unchanged.
44
44
 
45
45
  Credits: One Ciel LLC
46
46
 
47
- Current version: `0.1.29`
47
+ Current version: `0.1.30`
48
48
 
49
49
  ## Why This Exists
50
50
 
@@ -92,6 +92,40 @@ npm install -g @oneciel-ai/claude-any
92
92
  claude-any
93
93
  ```
94
94
 
95
+ ## Run Claude Code Headlessly
96
+
97
+ Use headless mode when a script, SSH session, CI job, or parent agent needs to
98
+ launch Claude Code directly without opening the pre-launch menu. `claude-any`
99
+ consumes `--ca-*` options, starts the required local router, and passes the
100
+ remaining arguments to Claude Code.
101
+
102
+ ```sh
103
+ claude-any --ca-provider nvidia-hosted --ca-model z-ai/glm-4.7
104
+ ```
105
+
106
+ ```sh
107
+ claude-any --ca-provider ollama-cloud --ca-model glm-5.1
108
+ ```
109
+
110
+ ```sh
111
+ claude-any --ca-provider ollama --ca-base-url http://127.0.0.1:11434 --ca-model qwen3-coder
112
+ ```
113
+
114
+ Run one non-interactive Claude Code prompt:
115
+
116
+ ```sh
117
+ claude-any --ca-provider nvidia-hosted --ca-model z-ai/glm-4.7 --ca-no-update-check -p "Reply with OK only." --output-format text
118
+ ```
119
+
120
+ Use the saved provider/model and only skip the menu:
121
+
122
+ ```sh
123
+ CLAUDE_ANY_SKIP_MENU=1 claude-any -p "Summarize this repository." --output-format text
124
+ ```
125
+
126
+ More examples are in [Headless Examples](#headless-examples) and
127
+ [the full manual](docs/manual.md#headless-usage).
128
+
95
129
  **Upgrade:**
96
130
 
97
131
  ```sh
@@ -300,17 +334,26 @@ steps under that larger model's supervision.
300
334
 
301
335
  ## Changelog
302
336
 
303
- ### 0.1.29
337
+ ### 0.1.30
304
338
 
305
- - **NVIDIA compatibility test fix**: `claude-any test` now restarts the local
306
- router before router-mode tests, so upgraded installs do not accidentally use
307
- an old long-running router that still expects `nvd-claude-proxy`.
308
- - **Clear NVIDIA router wording**: menu status now describes NVIDIA hosted as
309
- using the local claude-any router instead of the retired local proxy path.
339
+ - **Headless launch docs moved up**: README now shows copy-ready examples for
340
+ launching Claude Code directly with `--ca-provider`, `--ca-model`, `-p`, and
341
+ `CLAUDE_ANY_SKIP_MENU=1` immediately after install.
342
+ - **NVIDIA hosted wording cleanup**: provider and lifecycle docs now describe
343
+ NVIDIA hosted as using the Claude Any local router, with no separate hosted
344
+ API Catalog proxy requirement.
310
345
 
311
- ### 0.1.28
346
+ ### 0.1.29
312
347
 
313
- - **Plan Mode + Advisor headline**: document Claude Any's Plan Mode support for
348
+ - **NVIDIA compatibility test fix**: `claude-any test` now restarts the local
349
+ router before router-mode tests, so upgraded installs do not accidentally use
350
+ an old long-running router that still expects `nvd-claude-proxy`.
351
+ - **Clear NVIDIA router wording**: menu status now describes NVIDIA hosted as
352
+ using the local claude-any router instead of the retired local proxy path.
353
+
354
+ ### 0.1.28
355
+
356
+ - **Plan Mode + Advisor headline**: document Claude Any's Plan Mode support for
314
357
  router-backed non-Anthropic providers and the `/advisor` slash command backed
315
358
  by a selected long-context Advisor Model.
316
359
  - **Status-line RPM telemetry**: Claude Any installs a Claude Code `statusLine`
@@ -466,7 +509,7 @@ steps under that larger model's supervision.
466
509
  | Ollama | Native when available, router otherwise | Local Ollama normally needs no API key. Cloud models through local Ollama require `ollama signin` on the Ollama host. |
467
510
  | Ollama Cloud | Router | Calls `https://ollama.com/api`; requires an Ollama API key. |
468
511
  | vLLM | Native Anthropic-compatible endpoint | Use a vLLM endpoint that exposes Anthropic-compatible `/v1/messages`; match `--tool-call-parser` to the model family. |
469
- | NVIDIA hosted | Router/proxy | Uses NVIDIA hosted API through the compatibility path. |
512
+ | NVIDIA hosted | Router | Uses the NVIDIA hosted API Catalog through the Claude Any local router. |
470
513
  | self-hosted NIM | Native Anthropic-compatible endpoint | Use the self-hosted NIM Anthropic-compatible endpoint. |
471
514
 
472
515
  ## Service Lifecycle
@@ -474,17 +517,16 @@ steps under that larger model's supervision.
474
517
  Claude Any does not keep every possible backend helper running all the time. The
475
518
  normal lifecycle is:
476
519
 
477
- - Before launch, managed router/proxy processes can be stopped with
520
+ - Before launch, managed router processes can be stopped with
478
521
  `claude-any stop`.
479
522
  - When `claude-any` starts Claude Code, it starts only the services required by
480
523
  the selected provider.
481
524
  - Ollama and Ollama Cloud router mode use the Claude Any router on
482
525
  `127.0.0.1:8799`.
483
- - NVIDIA hosted router mode uses the Claude Any router on `127.0.0.1:8799` and
484
- starts `nvd-claude-proxy` on `127.0.0.1:8788` only when that provider needs it.
485
- - Switching away from NVIDIA hosted does not require keeping the NVIDIA proxy
486
- alive; stale sessions should be cleaned with `claude-any stop` before a fresh
487
- test or launch.
526
+ - NVIDIA hosted router mode uses the Claude Any router on `127.0.0.1:8799`;
527
+ hosted API Catalog models do not require a separate NVIDIA proxy.
528
+ - Run `claude-any stop` before a fresh provider-switch test if an old router is
529
+ still bound to the local port.
488
530
 
489
531
  This keeps Claude Code pointed at one stable Claude Any entry point while still
490
532
  letting provider-specific helpers start on demand.
@@ -512,7 +554,7 @@ vllm serve Qwen/Qwen3-Coder-30B-A3B-Instruct \
512
554
  ## Headless Examples
513
555
 
514
556
  Headless commands skip the pre-launch menu and launch Claude Code immediately.
515
- Claude Any consumes `--ca-*` setup flags, starts the required router/proxy
557
+ Claude Any consumes `--ca-*` setup flags, starts the required router
516
558
  services, then passes the remaining arguments to Claude Code.
517
559
 
518
560
  ```sh
package/claude_any.py CHANGED
@@ -84,7 +84,7 @@ PROVIDER_LABELS = {
84
84
  "self-hosted-nim": "Self Hosted NIM",
85
85
  }
86
86
  APP_NAME = "Claude Any"
87
- VERSION = "0.1.29"
87
+ VERSION = "0.1.30"
88
88
  CREDITS = "Credits: One Ciel LLC"
89
89
 
90
90
  LOG_LEVELS = {"SILENT": 0, "ERROR": 1, "WARN": 2, "INFO": 3, "DEBUG": 4, "TRACE": 5}
package/docs/README.ja.md CHANGED
@@ -43,7 +43,7 @@ vLLM、NVIDIA hosted、self-hosted NIM を選択し、通常の Claude Code 引
43
43
 
44
44
  Credits: One Ciel LLC
45
45
 
46
- 現在のバージョン: `0.1.29`
46
+ 現在のバージョン: `0.1.30`
47
47
 
48
48
  ## 作られた理由
49
49
 
@@ -87,6 +87,39 @@ npm install -g @oneciel-ai/claude-any
87
87
  claude-any
88
88
  ```
89
89
 
90
+ ## Claude Code を headless で直接起動
91
+
92
+ スクリプト、SSH セッション、CI ジョブ、親エージェントが事前メニューなしで
93
+ Claude Code を直接起動したい場合は headless mode を使います。
94
+ `claude-any` は `--ca-*` オプションを先に処理し、必要な local router を
95
+ 起動してから、残りの引数を Claude Code にそのまま渡します。
96
+
97
+ ```sh
98
+ claude-any --ca-provider nvidia-hosted --ca-model z-ai/glm-4.7
99
+ ```
100
+
101
+ ```sh
102
+ claude-any --ca-provider ollama-cloud --ca-model glm-5.1
103
+ ```
104
+
105
+ ```sh
106
+ claude-any --ca-provider ollama --ca-base-url http://127.0.0.1:11434 --ca-model qwen3-coder
107
+ ```
108
+
109
+ 1 回だけ実行する非対話 Claude Code prompt:
110
+
111
+ ```sh
112
+ claude-any --ca-provider nvidia-hosted --ca-model z-ai/glm-4.7 --ca-no-update-check -p "Reply with OK only." --output-format text
113
+ ```
114
+
115
+ 保存済み provider/model を使い、メニューだけをスキップ:
116
+
117
+ ```sh
118
+ CLAUDE_ANY_SKIP_MENU=1 claude-any -p "Summarize this repository." --output-format text
119
+ ```
120
+
121
+ その他の例は [manual](manual.md#headless-usage) を参照してください。
122
+
90
123
  **アップグレード:**
91
124
 
92
125
  ```sh
@@ -274,15 +307,24 @@ Windows/Linux 管理、クリーンアップスクリプト、定期的なセキ
274
307
 
275
308
  ## 変更履歴
276
309
 
310
+ ### 0.1.30
311
+
312
+ - **Headless 起動ドキュメントを上部へ移動**: インストール直後の README で
313
+ `--ca-provider`、`--ca-model`、`-p`、`CLAUDE_ANY_SKIP_MENU=1` を使って
314
+ Claude Code を直接起動する copy-ready な例を確認できます。
315
+ - **NVIDIA hosted の文言整理**: provider/lifecycle ドキュメントで NVIDIA
316
+ hosted を Claude Any local router 経由として説明し、hosted API Catalog に
317
+ 別 proxy が必要だという表現をなくしました。
318
+
277
319
  ### 0.1.29
278
320
 
279
321
  - **NVIDIA 互換性テストの修正**: `claude-any test` は router mode のテスト前に
280
- local router を再起動します。npm upgrade 後も古い常駐 router が
281
- `nvd-claude-proxy` を要求する問題を避けます。
282
- - **NVIDIA router 表示の整理**: メニューの状態表示を、廃止した local proxy
283
- 経路ではなく claude-any local router 基準に更新しました。
284
-
285
- ### 0.1.28
322
+ local router を再起動します。npm upgrade 後も古い常駐 router が
323
+ `nvd-claude-proxy` を要求する問題を避けます。
324
+ - **NVIDIA router 表示の整理**: メニューの状態表示を、廃止した local proxy
325
+ 経路ではなく claude-any local router 基準に更新しました。
326
+
327
+ ### 0.1.28
286
328
 
287
329
  - **Plan Mode + Advisor ヘッドライン**: router 経由の non-Anthropic provider での
288
330
  Plan Mode 対応と、選択した長コンテキスト Advisor Model で動作する `/advisor`
@@ -417,7 +459,7 @@ Windows/Linux 管理、クリーンアップスクリプト、定期的なセキ
417
459
  | Ollama | Native 優先、必要時 router | ローカル Ollama は通常 API key 不要。ローカル Ollama で `:cloud` model を使う場合は Ollama host で `ollama signin` が必要。 |
418
460
  | Ollama Cloud | Router | `https://ollama.com/api` を直接呼び出し、Ollama API key が必要。 |
419
461
  | vLLM | Native Anthropic-compatible endpoint | Anthropic 互換 `/v1/messages` endpoint を使い、モデル系列に合う `--tool-call-parser` を指定。 |
420
- | NVIDIA hosted | Router/proxy | NVIDIA hosted API を compatibility 経路で使用。 |
462
+ | NVIDIA hosted | Router | NVIDIA hosted API Catalog Claude Any local router 経由で使用。 |
421
463
  | self-hosted NIM | Native Anthropic-compatible endpoint | self-hosted NIM の Anthropic 互換 endpoint を使用。 |
422
464
 
423
465
  ## サービスライフサイクル
@@ -425,16 +467,14 @@ Windows/Linux 管理、クリーンアップスクリプト、定期的なセキ
425
467
  Claude Any は、すべての backend helper を常時起動しておく設計ではありません。
426
468
  通常のライフサイクルは次の通りです。
427
469
 
428
- - 起動前に、管理中の router/proxy は `claude-any stop` で停止できます。
470
+ - 起動前に、管理中の router は `claude-any stop` で停止できます。
429
471
  - `claude-any` が Claude Code を起動するとき、選択中 provider に必要な
430
472
  service だけを開始します。
431
473
  - Ollama/Ollama Cloud router mode は `127.0.0.1:8799` の Claude Any router
432
474
  を使います。
433
475
  - NVIDIA hosted router mode は `127.0.0.1:8799` の Claude Any router を使い、
434
- その provider に必要な場合だけ `127.0.0.1:8788` `nvd-claude-proxy`
435
- 開始します。
436
- - NVIDIA hosted から別 provider に切り替えるとき、NVIDIA proxy を生かし
437
- 続ける必要はありません。新しい test や launch の前に stale session は
476
+ hosted API Catalog model には別の NVIDIA proxy は不要です。
477
+ - provider 切り替えテスト前に古い router が local port を保持している場合は、
438
478
  `claude-any stop` で整理してください。
439
479
 
440
480
  この構成により、Claude Code は安定した Claude Any entry point を使いながら、
package/docs/README.ko.md CHANGED
@@ -43,7 +43,7 @@ NVIDIA hosted, self-hosted NIM을 선택하고, Claude Code의 일반 인자는
43
43
 
44
44
  Credits: One Ciel LLC
45
45
 
46
- 현재 버전: `0.1.29`
46
+ 현재 버전: `0.1.30`
47
47
 
48
48
  ## 왜 만들었나
49
49
 
@@ -87,6 +87,40 @@ npm install -g @oneciel-ai/claude-any
87
87
  claude-any
88
88
  ```
89
89
 
90
+ ## Claude Code를 headless로 바로 실행
91
+
92
+ 스크립트, SSH 세션, CI 작업, 상위 에이전트가 실행 전 메뉴 없이 Claude Code를
93
+ 바로 시작해야 할 때 headless mode를 사용합니다. `claude-any`는 `--ca-*`
94
+ 옵션을 먼저 소비하고, 필요한 local router를 시작한 뒤, 나머지 인자를 Claude
95
+ Code에 그대로 넘깁니다.
96
+
97
+ ```sh
98
+ claude-any --ca-provider nvidia-hosted --ca-model z-ai/glm-4.7
99
+ ```
100
+
101
+ ```sh
102
+ claude-any --ca-provider ollama-cloud --ca-model glm-5.1
103
+ ```
104
+
105
+ ```sh
106
+ claude-any --ca-provider ollama --ca-base-url http://127.0.0.1:11434 --ca-model qwen3-coder
107
+ ```
108
+
109
+ 한 번만 실행하는 비대화형 Claude Code prompt:
110
+
111
+ ```sh
112
+ claude-any --ca-provider nvidia-hosted --ca-model z-ai/glm-4.7 --ca-no-update-check -p "Reply with OK only." --output-format text
113
+ ```
114
+
115
+ 저장된 provider/model을 그대로 쓰고 메뉴만 건너뛰기:
116
+
117
+ ```sh
118
+ CLAUDE_ANY_SKIP_MENU=1 claude-any -p "Summarize this repository." --output-format text
119
+ ```
120
+
121
+ 더 많은 예시는 [헤드리스 예제](#헤드리스-에이전트-채팅)와
122
+ [전체 manual](manual.md#headless-usage)을 참고하세요.
123
+
90
124
  **업그레이드:**
91
125
 
92
126
  ```sh
@@ -273,17 +307,25 @@ Windows 이벤트 로그 리뷰, 바이러스/랜섬웨어 침입 시도 정리,
273
307
 
274
308
  ## 변경 이력
275
309
 
276
- ### 0.1.29
310
+ ### 0.1.30
277
311
 
278
- - **NVIDIA 호환성 테스트 수정**: `claude-any test`가 router mode 테스트 전에
279
- 로컬 router를 재시작하므로, 업그레이드 후에도 오래 떠 있던 예전 router가
280
- `nvd-claude-proxy`를 찾는 문제가 사라집니다.
281
- - **NVIDIA router 안내 정리**: 메뉴 상태 문구를 이상 사용하지 않는 local proxy
282
- 경로가 아니라 claude-any local router 기준으로 표시합니다.
312
+ - **Headless 실행 문서 상단 배치**: 설치 직후 README에서 `--ca-provider`,
313
+ `--ca-model`, `-p`, `CLAUDE_ANY_SKIP_MENU=1`로 Claude Code를 바로 실행하는
314
+ 복사용 예제를 수 있습니다.
315
+ - **NVIDIA hosted 문구 정리**: provider/lifecycle 문서가 NVIDIA hosted를 별도
316
+ hosted API Catalog proxy가 아닌 Claude Any local router 기준으로 설명합니다.
283
317
 
284
- ### 0.1.28
318
+ ### 0.1.29
285
319
 
286
- - **Plan Mode + Advisor 헤드라인**: router 기반 non-Anthropic provider 의
320
+ - **NVIDIA 호환성 테스트 수정**: `claude-any test`가 router mode 테스트 전에
321
+ 로컬 router를 재시작하므로, 업그레이드 후에도 오래 떠 있던 예전 router가
322
+ `nvd-claude-proxy`를 찾는 문제가 사라집니다.
323
+ - **NVIDIA router 안내 정리**: 메뉴 상태 문구를 더 이상 사용하지 않는 local proxy
324
+ 경로가 아니라 claude-any local router 기준으로 표시합니다.
325
+
326
+ ### 0.1.28
327
+
328
+ - **Plan Mode + Advisor 헤드라인**: router 기반 non-Anthropic provider 의
287
329
  Plan Mode 지원과, 선택한 긴 컨텍스트 Advisor Model 로 동작하는 `/advisor`
288
330
  slash command 를 문서화했습니다.
289
331
  - **statusLine RPM 표시**: Claude Any 가 Claude Code `statusLine` command 를
@@ -414,7 +456,7 @@ Windows 이벤트 로그 리뷰, 바이러스/랜섬웨어 침입 시도 정리,
414
456
  | Ollama | Native 우선, 필요 시 router | 로컬 Ollama는 보통 API 키가 필요 없습니다. 로컬 Ollama에서 `:cloud` 모델을 쓰려면 Ollama host에서 `ollama signin`이 필요합니다. |
415
457
  | Ollama Cloud | Router | `https://ollama.com/api` 직접 호출. Ollama API 키 필요. |
416
458
  | vLLM | Native Anthropic-compatible endpoint | Anthropic 호환 `/v1/messages`를 제공하는 vLLM endpoint 사용. 모델 계열에 맞는 `--tool-call-parser` 필요. |
417
- | NVIDIA hosted | Router/proxy | NVIDIA hosted API를 compatibility 경로로 사용. |
459
+ | NVIDIA hosted | Router | NVIDIA hosted API CatalogClaude Any local router로 사용. |
418
460
  | self-hosted NIM | Native Anthropic-compatible endpoint | self-hosted NIM Anthropic 호환 endpoint 사용. |
419
461
 
420
462
  ## 서비스 생명주기
@@ -422,17 +464,15 @@ Windows 이벤트 로그 리뷰, 바이러스/랜섬웨어 침입 시도 정리,
422
464
  Claude Any는 가능한 모든 backend helper를 항상 띄워두지 않습니다. 기본
423
465
  생명주기는 다음과 같습니다.
424
466
 
425
- - 실행 전 관리 중인 router/proxy는 `claude-any stop`으로 정리할 수 있습니다.
467
+ - 실행 전 관리 중인 router는 `claude-any stop`으로 정리할 수 있습니다.
426
468
  - `claude-any`가 Claude Code를 시작할 때, 선택된 provider에 필요한 서비스만
427
469
  시작합니다.
428
470
  - Ollama/Ollama Cloud router mode는 `127.0.0.1:8799`의 Claude Any router를
429
471
  사용합니다.
430
- - NVIDIA hosted router mode는 `127.0.0.1:8799`의 Claude Any router를 쓰고,
431
- 해당 provider에 필요할 때만 `127.0.0.1:8788`의 `nvd-claude-proxy`를
432
- 시작합니다.
433
- - NVIDIA hosted에서 다른 provider로 전환할 때 NVIDIA proxy를 계속 살려둘
434
- 필요는 없습니다. 새 테스트나 실행 전 stale session은 `claude-any stop`으로
435
- 정리하세요.
472
+ - NVIDIA hosted router mode는 `127.0.0.1:8799`의 Claude Any router를
473
+ 사용하며, hosted API Catalog 모델에는 별도 NVIDIA proxy가 필요 없습니다.
474
+ - provider 전환 테스트 전에 오래된 router가 local port를 잡고 있으면
475
+ `claude-any stop`으로 정리하세요.
436
476
 
437
477
  이 방식은 Claude Code가 하나의 안정적인 Claude Any 진입점을 사용하게 하면서,
438
478
  provider-specific helper는 필요한 시점에만 시작하도록 합니다.
package/docs/README.zh.md CHANGED
@@ -43,7 +43,7 @@ NIM,并把普通 Claude Code 参数原样传递。
43
43
 
44
44
  Credits: One Ciel LLC
45
45
 
46
- 当前版本: `0.1.29`
46
+ 当前版本: `0.1.30`
47
47
 
48
48
  ## 为什么存在
49
49
 
@@ -85,6 +85,38 @@ npm install -g @oneciel-ai/claude-any
85
85
  claude-any
86
86
  ```
87
87
 
88
+ ## Headless 直接启动 Claude Code
89
+
90
+ 当脚本、SSH 会话、CI 任务或上级 agent 需要跳过启动菜单并直接运行 Claude
91
+ Code 时,使用 headless mode。`claude-any` 会先消费 `--ca-*` 选项,启动所需
92
+ 的 local router,然后把剩余参数原样传给 Claude Code。
93
+
94
+ ```sh
95
+ claude-any --ca-provider nvidia-hosted --ca-model z-ai/glm-4.7
96
+ ```
97
+
98
+ ```sh
99
+ claude-any --ca-provider ollama-cloud --ca-model glm-5.1
100
+ ```
101
+
102
+ ```sh
103
+ claude-any --ca-provider ollama --ca-base-url http://127.0.0.1:11434 --ca-model qwen3-coder
104
+ ```
105
+
106
+ 执行一次非交互 Claude Code prompt:
107
+
108
+ ```sh
109
+ claude-any --ca-provider nvidia-hosted --ca-model z-ai/glm-4.7 --ca-no-update-check -p "Reply with OK only." --output-format text
110
+ ```
111
+
112
+ 使用已保存的 provider/model,只跳过菜单:
113
+
114
+ ```sh
115
+ CLAUDE_ANY_SKIP_MENU=1 claude-any -p "Summarize this repository." --output-format text
116
+ ```
117
+
118
+ 更多示例见 [manual](manual.md#headless-usage)。
119
+
88
120
  **升级:**
89
121
 
90
122
  ```sh
@@ -261,15 +293,24 @@ Hermes 格式模型或部分较旧的 Qwen tool template。
261
293
 
262
294
  ## 更新日志
263
295
 
296
+ ### 0.1.30
297
+
298
+ - **Headless 启动文档前置**:README 在安装后立即给出可复制示例,展示如何用
299
+ `--ca-provider`、`--ca-model`、`-p`、`CLAUDE_ANY_SKIP_MENU=1` 直接启动
300
+ Claude Code。
301
+ - **NVIDIA hosted 文案清理**:provider/lifecycle 文档现在说明 NVIDIA hosted
302
+ 通过 Claude Any local router 使用,不再暗示 hosted API Catalog 需要单独
303
+ proxy。
304
+
264
305
  ### 0.1.29
265
306
 
266
307
  - **NVIDIA 兼容性测试修复**:`claude-any test` 现在会在 router mode 测试前
267
- 重启本地 router,避免 npm upgrade 后仍连接到旧的常驻 router 并错误要求
268
- `nvd-claude-proxy`。
269
- - **NVIDIA router 状态文案更新**:菜单状态现在显示 claude-any local router,
270
- 不再提示已废弃的 local proxy 路径。
271
-
272
- ### 0.1.28
308
+ 重启本地 router,避免 npm upgrade 后仍连接到旧的常驻 router 并错误要求
309
+ `nvd-claude-proxy`。
310
+ - **NVIDIA router 状态文案更新**:菜单状态现在显示 claude-any local router,
311
+ 不再提示已废弃的 local proxy 路径。
312
+
313
+ ### 0.1.28
273
314
 
274
315
  - **Plan Mode + Advisor 标题**: 文档现在强调 router-backed non-Anthropic
275
316
  provider 的 Plan Mode 支持,以及由选定长上下文 Advisor Model 驱动的 `/advisor`
@@ -399,20 +440,20 @@ Hermes 格式模型或部分较旧的 Qwen tool template。
399
440
  | Ollama | Native 优先,必要时 router | 本地 Ollama 通常不需要 API key;通过本地 Ollama 使用 `:cloud` 模型时,需要在 Ollama host 上 `ollama signin`。 |
400
441
  | Ollama Cloud | Router | 直接调用 `https://ollama.com/api`,需要 Ollama API key。 |
401
442
  | vLLM | Native Anthropic-compatible endpoint | 使用 Anthropic 兼容 `/v1/messages` endpoint,并让 `--tool-call-parser` 匹配模型系列。 |
402
- | NVIDIA hosted | Router/proxy | 通过 compatibility 路径使用 NVIDIA hosted API。 |
443
+ | NVIDIA hosted | Router | 通过 Claude Any local router 使用 NVIDIA hosted API Catalog。 |
403
444
  | self-hosted NIM | Native Anthropic-compatible endpoint | 使用 self-hosted NIM 的 Anthropic 兼容 endpoint。 |
404
445
 
405
446
  ## 服务生命周期
406
447
 
407
448
  Claude Any 不会一直运行所有可能的 backend helper。正常生命周期如下:
408
449
 
409
- - 启动前,可用 `claude-any stop` 清理受管理的 router/proxy 进程。
450
+ - 启动前,可用 `claude-any stop` 清理受管理的 router 进程。
410
451
  - `claude-any` 启动 Claude Code 时,只启动当前所选 provider 需要的服务。
411
452
  - Ollama/Ollama Cloud router mode 使用 `127.0.0.1:8799` 上的 Claude Any router。
412
- - NVIDIA hosted router mode 使用 `127.0.0.1:8799` 上的 Claude Any router,并且只在
413
- provider 需要时启动 `127.0.0.1:8788` 上的 `nvd-claude-proxy`。
414
- - NVIDIA hosted 切换到其他 provider 时,不需要让 NVIDIA proxy 一直运行。新的
415
- test 或 launch 之前,请用 `claude-any stop` 清理 stale session。
453
+ - NVIDIA hosted router mode 使用 `127.0.0.1:8799` 上的 Claude Any router
454
+ hosted API Catalog 模型不需要单独的 NVIDIA proxy
455
+ - provider 切换测试前,如果旧 router 仍占用 local port,请用
456
+ `claude-any stop` 清理。
416
457
 
417
458
  这样 Claude Code 可以始终使用稳定的 Claude Any 入口,同时 provider-specific helper
418
459
  只在需要时启动。
package/docs/manual.md CHANGED
@@ -6,7 +6,7 @@ Code starts, while passing normal Claude Code arguments through unchanged.
6
6
 
7
7
  Credits: One Ciel LLC
8
8
 
9
- Current version: `0.1.29`
9
+ Current version: `0.1.30`
10
10
 
11
11
  ## Install
12
12
 
@@ -240,7 +240,7 @@ Headless mode is for launching Claude Code directly without opening the
240
240
  pre-launch menu. It is useful for SSH sessions, scripts, scheduled jobs,
241
241
  CI-like automation, and remote servers. Any `--ca-*` option updates Claude
242
242
  Any's saved configuration first, skips the menu, starts the required
243
- router/proxy services, then immediately executes Claude Code with the remaining
243
+ router services, then immediately executes Claude Code with the remaining
244
244
  arguments.
245
245
 
246
246
  Basic pattern:
@@ -300,7 +300,7 @@ claude-any \
300
300
  --ca-context-window 65536 \
301
301
  --ca-max-output-tokens 4096
302
302
 
303
- # NVIDIA hosted API Catalog through the local compatibility proxy
303
+ # NVIDIA hosted API Catalog through the local Claude Any router
304
304
  export NVIDIA_API_KEY="..."
305
305
  claude-any \
306
306
  --ca-provider nvidia-hosted \
@@ -367,7 +367,7 @@ Common Claude Any setup flags:
367
367
  | `--ca-disable-skills` / `--ca-enable-skills` | Control Claude Code skills for this launch. |
368
368
  | `--ca-no-update-check` | Skip the Claude Code update check. |
369
369
  | `--ca-status` | Print status and exit. |
370
- | `--ca-stop` | Stop managed router/proxy services and exit. |
370
+ | `--ca-stop` | Stop managed router services and exit. |
371
371
 
372
372
  Notes for automation:
373
373
 
@@ -502,15 +502,15 @@ agents can still perform web research through explicit tools.
502
502
  Claude Any starts provider helpers on demand rather than keeping every helper
503
503
  alive permanently.
504
504
 
505
- - `claude-any stop` stops managed Claude Any router and NVIDIA proxy processes.
505
+ - `claude-any stop` stops managed Claude Any router processes.
506
506
  - When launching Claude Code, Claude Any starts only the services required by
507
507
  the selected provider.
508
508
  - Ollama and Ollama Cloud router mode use the Claude Any router on
509
509
  `127.0.0.1:8799`.
510
- - NVIDIA hosted router mode uses the Claude Any router on `127.0.0.1:8799` and
511
- starts `nvd-claude-proxy` on `127.0.0.1:8788` only when that provider needs it.
510
+ - NVIDIA hosted router mode uses the Claude Any router on `127.0.0.1:8799`;
511
+ hosted API Catalog models do not require a separate NVIDIA proxy.
512
512
  - For clean provider-switch testing, run `claude-any stop`, select the provider,
513
- then launch or test. This avoids stale port ownership from old sessions.
513
+ then launch or test. This avoids stale router port ownership from old sessions.
514
514
 
515
515
  ## Development Story
516
516
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@oneciel-ai/claude-any",
3
- "version": "0.1.29",
3
+ "version": "0.1.30",
4
4
  "description": "Claude Code provider selector for Anthropic, Ollama, Ollama Cloud, vLLM, NVIDIA hosted, and self-hosted NIM.",
5
5
  "license": "MIT",
6
6
  "author": "One Ciel LLC",