dsh-local-ai 0.1.5 → 0.2.0

package/CHANGELOG.md

@@ -5,6 +5,12 @@ All notable changes to this project are documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.2.0] - 2026-08-26
+
+### Added
+
+- OpenAI-compatible multi-backend provider (LM Studio / vLLM / llama.cpp) with route-provider routing.
+
 ## [0.1.5] - 2026-08-23
 
 ### Added

package/README.es.md

@@ -109,8 +109,16 @@ Todos los ajustes son campos `Config` de Schemastery (modificables desde cordis.
 | `models[].contextWindow` | *(none)* | Capacidad de contexto por modelo |
 | `models[].maxTokens` | *(none)* | Límite de salida por modelo |
 | `models[].temperature` | *(none)* | Temperatura de muestreo por modelo |
+| `backends` | `[]` | Backends locales compatibles con OpenAI (LM Studio / vLLM / llama.cpp) |
+| `backends[].name` | *(required)* | Nombre del backend; registra el id de proveedor `openai:<name>` |
+| `backends[].baseURL` | *(required)* | URL base del backend (incluye `/v1`), p. ej. `http://127.0.0.1:1234/v1` |
+| `backends[].apiKey` | *(none)* | Clave bearer opcional (la mayoría de servidores locales la dejan vacía) |
+| `backends[].models` | `[]` | Mapeos de nombre visible → modelo del backend |
+| `backends[].maxTokens` | `4096` | Límite de salida por backend cuando un modelo no tiene valor exacto |
+| `backends[].temperature` | *(none)* | Temperatura de muestreo por backend |
 | `route` | `[]` | Reglas de enrutamiento a modelos locales (primera coincidencia gana) |
 | `route[].model` | *(required)* | Nombre del modelo local destino |
+| `route[].provider` | `ollama` | Id del proveedor destino: `ollama` u `openai:<name>` |
 | `route[].purpose` | *(none)* | Coincidencia de tipo de tarea: `compaction` / `session-title` |
 | `route[].keywords` | `[]` | Palabras clave de solicitud (sin distinguir mayúsculas) |
 | `route[].always` | `false` | Enruta toda solicitud elegible a este modelo |
@@ -171,6 +179,7 @@ pnpm pack           # el tarball publicado
 ## Contributors
 
 - [@PerryLink](https://github.com/PerryLink) — creador y mantenedor: adaptador, enrutamiento, herramientas, comprobación de estado, desinfección y documentación en cinco idiomas.
+- [@LABEST-IA](https://github.com/LABEST-IA) — corrección del CallId de las llamadas a herramientas (PR #2), y los informes sobre la ranura de llamadas a herramientas y el soporte de visión (issues #1, #3, #5).
 
 ## PerryLink DSH Plugin Family
 

package/README.hi.md

@@ -109,8 +109,16 @@ dsh --profile web --dump-config | grep -A2 'id: dsh-local-ai'
 | `models[].contextWindow` | *(none)* | प्रति-मॉडल संदर्भ क्षमता |
 | `models[].maxTokens` | *(none)* | प्रति-मॉडल आउटपुट सीमा |
 | `models[].temperature` | *(none)* | प्रति-मॉडल सैंपलिंग तापमान |
+| `backends` | `[]` | OpenAI-संगत स्थानीय बैकएंड (LM Studio / vLLM / llama.cpp) |
+| `backends[].name` | *(required)* | बैकएंड नाम; provider id `openai:<name>` पंजीकृत करता है |
+| `backends[].baseURL` | *(required)* | बैकएंड base URL (`/v1` सहित), जैसे `http://127.0.0.1:1234/v1` |
+| `backends[].apiKey` | *(none)* | वैकल्पिक bearer API key (अधिकांश स्थानीय सर्वर इसे खाली छोड़ते हैं) |
+| `backends[].models` | `[]` | दृश्य नाम → बैकएंड मॉडल मैपिंग |
+| `backends[].maxTokens` | `4096` | बिना सटीक मान वाले मॉडल के लिए प्रति-बैकएंड आउटपुट सीमा |
+| `backends[].temperature` | *(none)* | प्रति-बैकएंड सैंपलिंग तापमान |
 | `route` | `[]` | स्थानीय-मॉडल रूटिंग नियम (पहला मेल जीतता है) |
 | `route[].model` | *(required)* | लक्ष्य स्थानीय मॉडल नाम |
+| `route[].provider` | `ollama` | लक्ष्य प्रदाता id: `ollama` या `openai:<name>` |
 | `route[].purpose` | *(none)* | कार्य-प्रकार मेल: `compaction` / `session-title` |
 | `route[].keywords` | `[]` | केस-असंवेदी अनुरोध कीवर्ड |
 | `route[].always` | `false` | हर पात्र अनुरोध को इस मॉडल पर रूट करें |
@@ -171,6 +179,7 @@ pnpm pack           # प्रकाशित tarball
 ## Contributors
 
 - [@PerryLink](https://github.com/PerryLink) — निर्माता व अनुरक्षक: एडाप्टर, रूटिंग, टूल, स्वास्थ्य जाँच, सैनिटाइज़ेशन और पाँच-भाषा दस्तावेज़।
+- [@LABEST-IA](https://github.com/LABEST-IA) — टूल-कॉल CallId सुधार (PR #2), तथा टूल-कॉल स्लॉट व विज़न-समर्थन रिपोर्टें (issues #1, #3, #5)।
 
 ## PerryLink DSH Plugin Family
 

package/README.md

@@ -37,6 +37,7 @@
 - **Discovery & management** — `ollama_list` (installed models, running models, disk usage), `ollama_show` (parameter size, quantization, context length), `ollama_pull`, and `ollama_remove`.
 - **Health check** — process liveness (via the `ollama` CLI) and API responsiveness (via `/api/version`), reported as two independent signals.
 - **Official adapter** — the `ollama` provider route is registered through `ctx.llm.registerAdapter` (`LlmAdapter`), with configurable model mapping and temperature / max-tokens / stop translation.
+- **OpenAI-compatible backends** — LM Studio, vLLM, and llama.cpp `--server` each register as their own `openai:<name>` provider through the same `LlmAdapter` seam, reusing one OpenAI `/v1/chat/completions` adapter (text-only route).
 - **Local routing** — `model_route` rules route requests to a local model by task type (`purpose`), case-insensitive keyword, or `always`, with automatic fallback to the cloud when the local route fails before producing content.
 - **`/ollama` command** — a one-shot status overview: models, disk usage, health, and suggestions.
 - **Zero dependencies, HTTP first** — everything talks to Ollama's HTTP API (the CLI is used only for the process probe); no model files are bundled.
@@ -45,6 +46,7 @@
 request (loop)
    │ llm/stream waterfall
    ├─ rule matches? ──▶ route to ollama ──▶ Ollama /api/chat (NDJSON stream)
+   │              └─▶ route to openai:<name> ─▶ /v1/chat/completions (SSE)
    │                        └─ fails first ─▶ fall back to cloud (next())
    └─ no match ──▶ cloud provider
 tools ──▶ /api/tags · /api/ps · /api/show · /api/pull · /api/delete
@@ -110,8 +112,16 @@ All tunables are Schemastery `Config` fields (changeable from cordis.yml). An id
 | `models[].contextWindow` | *(none)* | Per-model context capacity |
 | `models[].maxTokens` | *(none)* | Per-model output cap |
 | `models[].temperature` | *(none)* | Per-model sampling temperature |
+| `backends` | `[]` | OpenAI-compatible local backends (LM Studio / vLLM / llama.cpp) |
+| `backends[].name` | *(required)* | Backend name; registers provider id `openai:<name>` |
+| `backends[].baseURL` | *(required)* | Backend base URL including `/v1`, e.g. `http://127.0.0.1:1234/v1` |
+| `backends[].apiKey` | *(none)* | Optional bearer API key (most local servers leave it empty) |
+| `backends[].models` | `[]` | Harness-visible → backend model mappings |
+| `backends[].maxTokens` | `4096` | Per-backend output cap used when a model has no exact value |
+| `backends[].temperature` | *(none)* | Per-backend sampling temperature |
 | `route` | `[]` | Local-model routing rules (first match wins) |
 | `route[].model` | *(required)* | Target local model name |
+| `route[].provider` | `ollama` | Target provider id: `ollama` or `openai:<name>` |
 | `route[].purpose` | *(none)* | Task type match: `compaction` / `session-title` |
 | `route[].keywords` | `[]` | Case-insensitive request keywords |
 | `route[].always` | `false` | Route every eligible request to this model |
@@ -172,6 +182,7 @@ pnpm pack           # the published tarball
 ## Contributors
 
 - [@PerryLink](https://github.com/PerryLink) — creator and maintainer: adapter, routing, tools, health check, sanitization, and the five-language docs.
+- [@LABEST-IA](https://github.com/LABEST-IA) — tool-call CallId fix (PR #2), and the tool-call slot and vision-support reports (issues #1, #3, #5).
 
 ## PerryLink DSH Plugin Family
 

package/README.pt.md

@@ -109,8 +109,16 @@ Todos os ajustes são campos `Config` de Schemastery (modificáveis pelo cordis.
 | `models[].contextWindow` | *(none)* | Capacidade de contexto por modelo |
 | `models[].maxTokens` | *(none)* | Limite de saída por modelo |
 | `models[].temperature` | *(none)* | Temperatura de amostragem por modelo |
+| `backends` | `[]` | Backends locais compatíveis com OpenAI (LM Studio / vLLM / llama.cpp) |
+| `backends[].name` | *(required)* | Nome do backend; registra o id de provedor `openai:<name>` |
+| `backends[].baseURL` | *(required)* | URL base do backend (inclui `/v1`), ex. `http://127.0.0.1:1234/v1` |
+| `backends[].apiKey` | *(none)* | Chave bearer opcional (a maioria dos servidores locais deixa vazia) |
+| `backends[].models` | `[]` | Mapeamentos de nome visível → modelo do backend |
+| `backends[].maxTokens` | `4096` | Limite de saída por backend quando um modelo não tem valor exato |
+| `backends[].temperature` | *(none)* | Temperatura de amostragem por backend |
 | `route` | `[]` | Regras de roteamento para modelos locais (primeira correspondência vence) |
 | `route[].model` | *(required)* | Nome do modelo local de destino |
+| `route[].provider` | `ollama` | Id do provedor de destino: `ollama` ou `openai:<name>` |
 | `route[].purpose` | *(none)* | Correspondência de tipo de tarefa: `compaction` / `session-title` |
 | `route[].keywords` | `[]` | Palavras-chave da solicitação (sem diferenciar maiúsculas) |
 | `route[].always` | `false` | Roteia toda solicitação elegível para este modelo |
@@ -171,6 +179,7 @@ pnpm pack           # o tarball publicado
 ## Contributors
 
 - [@PerryLink](https://github.com/PerryLink) — criador e mantenedor: adaptador, roteamento, ferramentas, verificação de saúde, sanitização e documentação em cinco idiomas.
+- [@LABEST-IA](https://github.com/LABEST-IA) — correção do CallId de chamadas de ferramenta (PR #2), e os relatórios sobre o slot de chamadas de ferramenta e o suporte a visão (issues #1, #3, #5).
 
 ## PerryLink DSH Plugin Family
 

package/README.zh.md

@@ -109,8 +109,16 @@ dsh --profile web --dump-config | grep -A2 'id: dsh-local-ai'
 | `models[].contextWindow` | *(none)* | 该模型的上下文容量 |
 | `models[].maxTokens` | *(none)* | 该模型的输出上限 |
 | `models[].temperature` | *(none)* | 该模型的采样温度 |
+| `backends` | `[]` | OpenAI 兼容的本地后端（LM Studio / vLLM / llama.cpp） |
+| `backends[].name` | *(required)* | 后端名；注册为 provider id `openai:<name>` |
+| `backends[].baseURL` | *(required)* | 后端 base URL（含 `/v1`），如 `http://127.0.0.1:1234/v1` |
+| `backends[].apiKey` | *(none)* | 可选 bearer API key（多数本地服务器留空） |
+| `backends[].models` | `[]` | 界面可见名 → 后端模型映射 |
+| `backends[].maxTokens` | `4096` | 该后端无精确值时的输出上限 |
+| `backends[].temperature` | *(none)* | 该后端的采样温度 |
 | `route` | `[]` | 本地模型路由规则（首个命中生效） |
 | `route[].model` | *(required)* | 目标本地模型名 |
+| `route[].provider` | `ollama` | 目标 provider id：`ollama` 或 `openai:<name>` |
 | `route[].purpose` | *(none)* | 任务类型匹配：`compaction` / `session-title` |
 | `route[].keywords` | `[]` | 不区分大小写的请求关键词 |
 | `route[].always` | `false` | 把所有符合条件的请求路由到此模型 |
@@ -171,6 +179,7 @@ pnpm pack           # 发布用 tarball
 ## Contributors
 
 - [@PerryLink](https://github.com/PerryLink) — 创建者与维护者：适配器、路由、工具、健康检查、脱敏与五语文档。
+- [@LABEST-IA](https://github.com/LABEST-IA) — 工具调用 CallId 修复（PR #2），以及工具调用槽位与视觉支持的报告（issue #1、#3、#5）。
 
 ## PerryLink DSH Plugin Family
 

package/cordis.patch.yml

@@ -32,15 +32,27 @@
         #     contextWindow: 4096
         #     maxTokens: 2048
         #     temperature: 0.7
+        # OpenAI-compatible backends (LM Studio / vLLM / llama.cpp). Each
+        # registers as provider id "openai:<name>"; baseURL includes the /v1
+        # prefix; apiKey is optional (most local servers leave it empty).
+        backends: []
+        # backends:
+        #   - name: lmstudio
+        #     baseURL: http://127.0.0.1:1234/v1
+        #   - name: vllm
+        #     baseURL: http://127.0.0.1:8000/v1
+        #     apiKey: token-abc123
         # Routing rules (first match wins): route requests to a local model by
         # task type (purpose), keyword (case-insensitive), or always. A rule
         # needs at least one of purpose/keywords/always; an empty list means
-        # no automatic re-routing.
+        # no automatic re-routing. `provider` defaults to "ollama"; use
+        # "openai:<name>" to target a backend above.
         route: []
         # route:
         #   - model: local
         #     keywords: ["confidential", "offline", "internal"]
         #   - model: local
         #     purpose: compaction
-        #   - model: local
+        #   - model: qwen2.5-7b
+        #     provider: openai:lmstudio
         #     always: true   # offline-first: route everything local