@oneciel-ai/claude-any 0.1.28 → 0.1.29

package/README.md

@@ -1,9 +1,9 @@
-# Claude Any
-
-![Claude Any: full Claude Code experience with free or low-cost LLMs](claude-any-adv.png)
-
-| English | [한국어](docs/README.ko.md) | [日本語](docs/README.ja.md) | [中文](docs/README.zh.md) |
-| --- | --- | --- | --- |
+# Claude Any
+
+![Claude Any: full Claude Code experience with free or low-cost LLMs](claude-any-adv.png)
+
+| English | [한국어](docs/README.ko.md) | [日本語](docs/README.ja.md) | [中文](docs/README.zh.md) |
+| --- | --- | --- | --- |
 
 [![npm version](https://img.shields.io/npm/v/@oneciel-ai/claude-any?logo=npm&label=npm)](https://www.npmjs.com/package/@oneciel-ai/claude-any)
 [![npm downloads](https://img.shields.io/npm/dm/@oneciel-ai/claude-any?logo=npm&label=downloads)](https://www.npmjs.com/package/@oneciel-ai/claude-any)
@@ -11,23 +11,23 @@
 
 > ## 🚀 Use the full Claude Code experience with free or low-cost LLMs
 >
-> - **Free** — [NVIDIA hosted NIM](https://build.nvidia.com/) (qwen3-coder-480b, gpt-oss, and friends) through the API Catalog.
-> - **Low-cost** — [Ollama Cloud](https://ollama.com/cloud) for GLM, Qwen, DeepSeek, and other open-weight models at a fraction of frontier-model pricing.
-> - **Free + local** — [Ollama](https://ollama.com/) or [vLLM](https://github.com/vllm-project/vllm) on your own GPU, fully offline.
-> - **Plan Mode + Advisor ready** — Claude Any preserves Claude Code Plan Mode on non-Anthropic providers and adds an optional long-context Advisor model for review.
-> - **Smooth free-model pacing** — Claude Code spends time reading files and running tools, and Claude Any uses that natural gap for RPM pacing so NVIDIA hosted free models feel usable even with strict per-minute limits.
->
-> Provider, model, base URL, API key, streaming behavior, and LLM options are all selected from a console menu **before** Claude Code starts. Claude Code itself runs untouched with all of its native tooling, slash commands, and workflows.
-
-## Today's Top 3 Benefits
-
-1. **Plan Mode works on non-Anthropic models** — Claude Any keeps Claude Code's Plan Mode usable even when the upstream provider is NVIDIA hosted, Ollama Cloud, local Ollama, vLLM, or NIM.
-2. **Advisor review with a bigger model** — pick a long-context Advisor Model at launch, then use `/advisor` inside Claude Code to review the current task, blockers, and next concrete action.
-3. **Free-model RPM limits feel smoother** — router-side RPM pacing uses the natural time spent reading files and running tools, so NVIDIA hosted free models can stay within per-minute limits with less visible waiting.
-
-### Demo
-
-![NVIDIA hosted NIM driving Claude Code (deepseek-4-flash)](docs/assets/claude-any-nvidia-nim.gif)
+> - **Free** — [NVIDIA hosted NIM](https://build.nvidia.com/) (qwen3-coder-480b, gpt-oss, and friends) through the API Catalog.
+> - **Low-cost** — [Ollama Cloud](https://ollama.com/cloud) for GLM, Qwen, DeepSeek, and other open-weight models at a fraction of frontier-model pricing.
+> - **Free + local** — [Ollama](https://ollama.com/) or [vLLM](https://github.com/vllm-project/vllm) on your own GPU, fully offline.
+> - **Plan Mode + Advisor ready** — Claude Any preserves Claude Code Plan Mode on non-Anthropic providers and adds an optional long-context Advisor model for review.
+> - **Smooth free-model pacing** — Claude Code spends time reading files and running tools, and Claude Any uses that natural gap for RPM pacing so NVIDIA hosted free models feel usable even with strict per-minute limits.
+>
+> Provider, model, base URL, API key, streaming behavior, and LLM options are all selected from a console menu **before** Claude Code starts. Claude Code itself runs untouched with all of its native tooling, slash commands, and workflows.
+
+## Today's Top 3 Benefits
+
+1. **Plan Mode works on non-Anthropic models** — Claude Any keeps Claude Code's Plan Mode usable even when the upstream provider is NVIDIA hosted, Ollama Cloud, local Ollama, vLLM, or NIM.
+2. **Advisor review with a bigger model** — pick a long-context Advisor Model at launch, then use `/advisor` inside Claude Code to review the current task, blockers, and next concrete action.
+3. **Free-model RPM limits feel smoother** — router-side RPM pacing uses the natural time spent reading files and running tools, so NVIDIA hosted free models can stay within per-minute limits with less visible waiting.
+
+### Demo
+
+![NVIDIA hosted NIM driving Claude Code (deepseek-4-flash)](docs/assets/claude-any-nvidia-nim.gif)
 
 NVIDIA hosted NIM (deepseek-4-flash) driving Claude Code through the claude-any router.  [full mp4 ⤓](https://github.com/OneCielAI/claude-any/raw/main/demo/claude-any-nvidia-nim.mp4)
 
@@ -44,7 +44,7 @@ arguments through unchanged.
 
 Credits: One Ciel LLC
 
-Current version: `0.1.28`
+Current version: `0.1.29`
 
 ## Why This Exists
 
@@ -84,23 +84,23 @@ Requirements:
 
 **Install from the npm registry (recommended):**
 
-```sh
-npm install -g @oneciel-ai/claude-any
-```
-
-```sh
-claude-any
-```
+```sh
+npm install -g @oneciel-ai/claude-any
+```
+
+```sh
+claude-any
+```
 
 **Upgrade:**
 
-```sh
-npm update -g @oneciel-ai/claude-any
-```
-
-```sh
-claude-any version
-```
+```sh
+npm update -g @oneciel-ai/claude-any
+```
+
+```sh
+claude-any version
+```
 
 **Uninstall:**
 
@@ -113,49 +113,49 @@ npm uninstall -g @oneciel-ai/claude-any
 Install directly from the GitHub repository (useful for testing unreleased
 commits between npm publishes):
 
-```sh
-npm install -g https://github.com/OneCielAI/claude-any.git
-```
-
-```sh
-claude-any
-```
+```sh
+npm install -g https://github.com/OneCielAI/claude-any.git
+```
+
+```sh
+claude-any
+```
 
 POSIX source install:
 
-```sh
-git clone https://github.com/OneCielAI/claude-any.git
-```
-
-```sh
-cd claude-any
-```
-
-```sh
-./install.sh
-```
-
-```sh
-claude-any
-```
+```sh
+git clone https://github.com/OneCielAI/claude-any.git
+```
+
+```sh
+cd claude-any
+```
+
+```sh
+./install.sh
+```
+
+```sh
+claude-any
+```
 
 Windows PowerShell source install:
 
-```powershell
-git clone https://github.com/OneCielAI/claude-any.git
-```
-
-```powershell
-cd claude-any
-```
-
-```powershell
-.\install.ps1
-```
-
-```powershell
-claude-any
-```
+```powershell
+git clone https://github.com/OneCielAI/claude-any.git
+```
+
+```powershell
+cd claude-any
+```
+
+```powershell
+.\install.ps1
+```
+
+```powershell
+claude-any
+```
 
 ### Releasing (maintainers)
 
@@ -263,24 +263,24 @@ steps under that larger model's supervision.
 - Native paths where providers expose Claude/Anthropic-compatible endpoints.
 - Router mode for providers that need request/response adaptation.
 - DuckDuckGo and fetch MCP wiring for non-native providers.
-- Headless setup flags such as `--ca-provider`, `--ca-model`, `--ca-base-url`,
-  `--ca-api-key-env`, `--ca-ollama-option`, and `--ca-max-output-tokens`.
-- Claude Code Plan Mode support on router-backed non-Anthropic providers,
-  including local handling for `EnterPlanMode` and plan artifacts.
-- Optional `/advisor` slash command that routes the current task state to a
-  selected Advisor Model, useful for long-context review and next-step checks.
-- Claude Code `statusLine` integration showing router RPM usage and wait time
-  in the bottom status area instead of polluting the chat transcript.
-- Router-side RPM control for NVIDIA hosted, self-hosted NIM, Ollama, and
-  Ollama Cloud. `rate_limit_rpm=0` disables throttling while still showing the
-  last-60-seconds usage rate.
-- Soft pacing subtracts time already spent reading files, running commands, and
-  waiting for tool results. In real coding sessions, those tool-call gaps absorb
-  much of the RPM spacing naturally, so providers such as NVIDIA hosted NIM can
-  stay within free-model limits without making every Claude Code turn feel
-  rate-limited.
-- Streaming proxy for Ollama/Ollama Cloud router path — tokens are delivered
-  to Claude Code as they arrive instead of waiting for the full response.
+- Headless setup flags such as `--ca-provider`, `--ca-model`, `--ca-base-url`,
+  `--ca-api-key-env`, `--ca-ollama-option`, and `--ca-max-output-tokens`.
+- Claude Code Plan Mode support on router-backed non-Anthropic providers,
+  including local handling for `EnterPlanMode` and plan artifacts.
+- Optional `/advisor` slash command that routes the current task state to a
+  selected Advisor Model, useful for long-context review and next-step checks.
+- Claude Code `statusLine` integration showing router RPM usage and wait time
+  in the bottom status area instead of polluting the chat transcript.
+- Router-side RPM control for NVIDIA hosted, self-hosted NIM, Ollama, and
+  Ollama Cloud. `rate_limit_rpm=0` disables throttling while still showing the
+  last-60-seconds usage rate.
+- Soft pacing subtracts time already spent reading files, running commands, and
+  waiting for tool results. In real coding sessions, those tool-call gaps absorb
+  much of the RPM spacing naturally, so providers such as NVIDIA hosted NIM can
+  stay within free-model limits without making every Claude Code turn feel
+  rate-limited.
+- Streaming proxy for Ollama/Ollama Cloud router path — tokens are delivered
+  to Claude Code as they arrive instead of waiting for the full response.
 - Per-provider `stream` on/off toggle and `stream_word_chunking` option to
   batch text deltas at word boundaries, mitigating SSE fragmentation that can
   break tool-call / JSON parsing in long streamed responses.
@@ -292,58 +292,66 @@ steps under that larger model's supervision.
   including `WorktreeCreate` / `WorktreeRemove`, so non-git working directories
   no longer fail Agent isolation with
   `Cannot create agent worktree: not in a git repository...`.
-- Config file caching — settings are read from disk once and reused until the
-  file changes, reducing per-request overhead in the router.
-- Router control-plane endpoints for headless agent coordination:
-  `/ca/chat/messages`, `/ca/chat/wait`, `/ca/chat/stream`, `/ca/chat/files`,
-  and `/ca/plan/artifacts`.
+- Config file caching — settings are read from disk once and reused until the
+  file changes, reducing per-request overhead in the router.
+- Router control-plane endpoints for headless agent coordination:
+  `/ca/chat/messages`, `/ca/chat/wait`, `/ca/chat/stream`, `/ca/chat/files`,
+  and `/ca/plan/artifacts`.
 
 ## Changelog
 
-### 0.1.28
-
-- **Plan Mode + Advisor headline**: document Claude Any's Plan Mode support for
-  router-backed non-Anthropic providers and the `/advisor` slash command backed
-  by a selected long-context Advisor Model.
-- **Status-line RPM telemetry**: Claude Any installs a Claude Code `statusLine`
-  command that shows router RPM usage and the latest wait time in the bottom
-  status area, keeping rate-limit telemetry out of the chat transcript.
-- **Soft RPM pacing for free hosted models**: NVIDIA hosted, self-hosted NIM,
-  Ollama, and Ollama Cloud can use router-side RPM pacing. The pacing subtracts
-  time already spent in file reads, command execution, and tool-result waits, so
-  normal coding tool-call gaps naturally absorb much of the RPM spacing.
-- **Unlimited usage display**: `rate_limit_rpm=0` disables throttling while
-  still displaying the last-60-seconds request rate.
-
-### 0.1.27
-
-- **Plan mode support for non-Anthropic providers**: the router now keeps
-  `EnterPlanMode` available and supports Claude Code Plan mode even when the
-  upstream model does not reliably choose that internal tool. Forced
-  `tool_choice=EnterPlanMode` is answered locally with a valid Anthropic
-  `tool_use`, and long implementation requests that receive only a short or
-  empty non-actionable text response are promoted to `EnterPlanMode` using
-  language-agnostic structure checks.
-- **Plan-mode self-tool handling**: unsupported Claude Code self-tools are
-  still stripped for non-Anthropic providers, but Plan-mode tools are handled
-  separately so planning can work instead of being disabled.
+### 0.1.29
 
-### 0.1.25
+- **NVIDIA compatibility test fix**: `claude-any test` now restarts the local
+  router before router-mode tests, so upgraded installs do not accidentally use
+  an old long-running router that still expects `nvd-claude-proxy`.
+- **Clear NVIDIA router wording**: menu status now describes NVIDIA hosted as
+  using the local claude-any router instead of the retired local proxy path.
 
-- **Plan-mode diagnostics**: set `~/.config/claude-any/log-level` to `TRACE`
-  to capture redacted request and response summaries in `requests.jsonl` /
-  `responses.jsonl`.
-- **Headless agent chat service**: the router exposes a small HTTP control
-  plane for sub coding agents. Agents can post messages, poll updates after
-  the last seen message id, or wait on an SSE stream when they do not have
-  their own loop.
-- **Plan artifact serving**: agents can create plan files through the router
-  and share stable local URLs, matching Claude Code's file/artifact-oriented
-  Plan-mode workflow without copying Anthropic's internal implementation.
-
-### 0.1.24
+### 0.1.28
 
-- **First public npm release** under the correct scope: `@oneciel-ai/claude-any`. Earlier 0.1.x versions were never published to the registry; this is the version that is actually installable via `npm install -g @oneciel-ai/claude-any`.
+- **Plan Mode + Advisor headline**: document Claude Any's Plan Mode support for
+  router-backed non-Anthropic providers and the `/advisor` slash command backed
+  by a selected long-context Advisor Model.
+- **Status-line RPM telemetry**: Claude Any installs a Claude Code `statusLine`
+  command that shows router RPM usage and the latest wait time in the bottom
+  status area, keeping rate-limit telemetry out of the chat transcript.
+- **Soft RPM pacing for free hosted models**: NVIDIA hosted, self-hosted NIM,
+  Ollama, and Ollama Cloud can use router-side RPM pacing. The pacing subtracts
+  time already spent in file reads, command execution, and tool-result waits, so
+  normal coding tool-call gaps naturally absorb much of the RPM spacing.
+- **Unlimited usage display**: `rate_limit_rpm=0` disables throttling while
+  still displaying the last-60-seconds request rate.
+
+### 0.1.27
+
+- **Plan mode support for non-Anthropic providers**: the router now keeps
+  `EnterPlanMode` available and supports Claude Code Plan mode even when the
+  upstream model does not reliably choose that internal tool. Forced
+  `tool_choice=EnterPlanMode` is answered locally with a valid Anthropic
+  `tool_use`, and long implementation requests that receive only a short or
+  empty non-actionable text response are promoted to `EnterPlanMode` using
+  language-agnostic structure checks.
+- **Plan-mode self-tool handling**: unsupported Claude Code self-tools are
+  still stripped for non-Anthropic providers, but Plan-mode tools are handled
+  separately so planning can work instead of being disabled.
+
+### 0.1.25
+
+- **Plan-mode diagnostics**: set `~/.config/claude-any/log-level` to `TRACE`
+  to capture redacted request and response summaries in `requests.jsonl` /
+  `responses.jsonl`.
+- **Headless agent chat service**: the router exposes a small HTTP control
+  plane for sub coding agents. Agents can post messages, poll updates after
+  the last seen message id, or wait on an SSE stream when they do not have
+  their own loop.
+- **Plan artifact serving**: agents can create plan files through the router
+  and share stable local URLs, matching Claude Code's file/artifact-oriented
+  Plan-mode workflow without copying Anthropic's internal implementation.
+
+### 0.1.24
+
+- **First public npm release** under the correct scope: `@oneciel-ai/claude-any`. Earlier 0.1.x versions were never published to the registry; this is the version that is actually installable via `npm install -g @oneciel-ai/claude-any`.
 
 ### 0.1.23
 
@@ -515,32 +523,32 @@ claude-any --ca-provider vllm --ca-base-url http://127.0.0.1:8000 --ca-model Qwe
 claude-any --ca-no-update-check -p "Reply with OK only." --output-format text
 ```
 
-All other arguments are passed through to Claude Code.
-
-## Headless Agent Chat
-
-When the claude-any router is running, sub agents can coordinate through local
-HTTP endpoints without opening the menu:
-
-```sh
-# Send a message to a channel.
-curl -s http://127.0.0.1:8799/ca/chat/messages \
-  -H 'content-type: application/json' \
-  -d '{"channel":"agents","sender_id":"codex","recipients":["kimi"],"message":"Need logs after id 42"}'
-
-# Poll updates after the last message id.
-curl -s 'http://127.0.0.1:8799/ca/chat/messages?channel=agents&recipient=kimi&after=42'
-
-# Wait on a stream until messages arrive.
-curl -N 'http://127.0.0.1:8799/ca/chat/stream?channel=agents&recipient=kimi&after=42&timeout=300'
-
-# Publish a plan file and get a served URL.
-curl -s http://127.0.0.1:8799/ca/plan/artifacts \
-  -H 'content-type: application/json' \
-  -d '{"title":"handoff","name":"handoff.md","content":"# Plan\n- step 1"}'
-```
-
-## Security
+All other arguments are passed through to Claude Code.
+
+## Headless Agent Chat
+
+When the claude-any router is running, sub agents can coordinate through local
+HTTP endpoints without opening the menu:
+
+```sh
+# Send a message to a channel.
+curl -s http://127.0.0.1:8799/ca/chat/messages \
+  -H 'content-type: application/json' \
+  -d '{"channel":"agents","sender_id":"codex","recipients":["kimi"],"message":"Need logs after id 42"}'
+
+# Poll updates after the last message id.
+curl -s 'http://127.0.0.1:8799/ca/chat/messages?channel=agents&recipient=kimi&after=42'
+
+# Wait on a stream until messages arrive.
+curl -N 'http://127.0.0.1:8799/ca/chat/stream?channel=agents&recipient=kimi&after=42&timeout=300'
+
+# Publish a plan file and get a served URL.
+curl -s http://127.0.0.1:8799/ca/plan/artifacts \
+  -H 'content-type: application/json' \
+  -d '{"title":"handoff","name":"handoff.md","content":"# Plan\n- step 1"}'
+```
+
+## Security
 
 Do not commit runtime configuration or API keys. Claude Any stores local runtime
 configuration under `~/.config/claude-any/`. NVIDIA hosted credentials used by

package/claude-any-menu.py

@@ -656,7 +656,7 @@ def probe_base_url(provider: str, pcfg: dict) -> str:
     if "your-" in base:
         return f"Base URL: placeholder ({base})"
     if provider == "nvidia-hosted":
-        return f"Base URL: NVIDIA hosted ({base}); proxy starts on launch"
+        return f"Base URL: NVIDIA hosted ({base}); local router http://127.0.0.1:8799 starts on launch"
     path = "/api/tags" if provider in ("ollama", "ollama-cloud") else "/v1/models"
     url = join_url(base, path)
     headers = {}