updating configuration tree view (#1088)

* simplify apis * show models * missing docs * fix model listing * handle no-url * don't sort * show errors * properly resolve fallbackt ools * add english prompt * use repo config * add dispatch * normalize configs * updated demo prompts
microsoft · Feb 2, 2025 · 1d4b443 · 1d4b443
1 parent 10279ca
commit 1d4b443
Show file tree

Hide file tree

Showing 30 changed files with 215 additions and 186 deletions.
diff --git a/.github/workflows/genai-blog-post.yml b/.github/workflows/genai-blog-post.yml
@@ -4,6 +4,12 @@ on:
 concurrency:
     group: ${{ github.workflow }}-${{ github.ref }}
     cancel-in-progress: true
+env:
+    GENAISCRIPT_DEFAULT_REASONING_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_MODEL }}
+    GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_MODEL }}
+    GENAISCRIPT_DEFAULT_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_VISION_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_VISION_MODEL }}
 jobs:
     build:
         runs-on: ubuntu-latest

diff --git a/.github/workflows/genai-commander.yml b/.github/workflows/genai-commander.yml
@@ -3,11 +3,11 @@ on:
     issue_comment:
         types: [created]
 env:
-    # Configure default GenAIScript models
-    # using Ollama's models
-    GENAISCRIPT_DEFAULT_MODEL: ollama:qwen2.5-coder:7b
-    GENAISCRIPT_DEFAULT_SMALL_MODEL: ollama:qwen2.5-coder:1.5b
-    GENAISCRIPT_DEFAULT_VISION_MODEL: ollama:llama3.2-vision:11b
+    GENAISCRIPT_DEFAULT_REASONING_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_MODEL }}
+    GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_MODEL }}
+    GENAISCRIPT_DEFAULT_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_VISION_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_VISION_MODEL }}
 jobs:
     pr_commented:
         # must be PR and have a comment starting with /genai

diff --git a/.github/workflows/genai-investigator.yml b/.github/workflows/genai-investigator.yml
@@ -12,11 +12,11 @@ permissions:
     actions: read
     pull-requests: write
 env:
-    # Configure default GenAIScript models
-    # using Ollama's models
-    GENAISCRIPT_DEFAULT_MODEL: ollama:qwen2.5-coder:7b
-    GENAISCRIPT_DEFAULT_SMALL_MODEL: ollama:qwen2.5-coder:1.5b
-    GENAISCRIPT_DEFAULT_VISION_MODEL: ollama:llama3.2-vision:11b
+    GENAISCRIPT_DEFAULT_REASONING_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_MODEL }}
+    GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_MODEL }}
+    GENAISCRIPT_DEFAULT_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_VISION_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_VISION_MODEL }}
 jobs:
     investigate:
         # Only run this job if the workflow run concluded with a failure

diff --git a/.github/workflows/genai-issue-review.yml b/.github/workflows/genai-issue-review.yml
@@ -6,11 +6,11 @@ concurrency:
     group: issues-${{ github.event.issue.number }}
     cancel-in-progress: true
 env:
-    # Configure default GenAIScript models
-    # using Ollama's models
-    GENAISCRIPT_DEFAULT_MODEL: ollama:qwen2.5-coder:7b
-    GENAISCRIPT_DEFAULT_SMALL_MODEL: ollama:qwen2.5-coder:1.5b
-    GENAISCRIPT_DEFAULT_VISION_MODEL: ollama:llama3.2-vision:11b
+    GENAISCRIPT_DEFAULT_REASONING_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_MODEL }}
+    GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_MODEL }}
+    GENAISCRIPT_DEFAULT_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_VISION_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_VISION_MODEL }}
 jobs:
     build:
         runs-on: ubuntu-latest

diff --git a/.github/workflows/genai-pr-commit-review.yml b/.github/workflows/genai-pr-commit-review.yml
@@ -8,11 +8,11 @@ on:
             - "packages/cli/**/*"
             - "packages/samples/**/*"
 env:
-    # Configure default GenAIScript models
-    # using Ollama's models
-    GENAISCRIPT_DEFAULT_MODEL: ollama:qwen2.5-coder:7b
-    GENAISCRIPT_DEFAULT_SMALL_MODEL: ollama:qwen2.5-coder:1.5b
-    GENAISCRIPT_DEFAULT_VISION_MODEL: ollama:llama3.2-vision:11b
+    GENAISCRIPT_DEFAULT_REASONING_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_MODEL }}
+    GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_MODEL }}
+    GENAISCRIPT_DEFAULT_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_VISION_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_VISION_MODEL }}
 jobs:
     build:
         runs-on: ubuntu-latest

diff --git a/.github/workflows/genai-pr-docs-commit-review.yml b/.github/workflows/genai-pr-docs-commit-review.yml
@@ -5,11 +5,11 @@ on:
             - docs/**/*.md
             - docs/**/*.mdx
 env:
-    # Configure default GenAIScript models
-    # using Ollama's models
-    GENAISCRIPT_DEFAULT_MODEL: ollama:qwen2.5-coder:7b
-    GENAISCRIPT_DEFAULT_SMALL_MODEL: ollama:qwen2.5-coder:1.5b
-    GENAISCRIPT_DEFAULT_VISION_MODEL: ollama:llama3.2-vision:11b
+    GENAISCRIPT_DEFAULT_REASONING_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_MODEL }}
+    GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_MODEL }}
+    GENAISCRIPT_DEFAULT_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_VISION_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_VISION_MODEL }}
 jobs:
     build:
         runs-on: ubuntu-latest

diff --git a/.github/workflows/genai-pr-review.yml b/.github/workflows/genai-pr-review.yml
@@ -1,5 +1,6 @@
 name: genai pull request review
 on:
+    workflow_dispatch:
     pull_request:
         types: [opened, ready_for_review, reopened]
         paths:
@@ -12,11 +13,11 @@ concurrency:
     group: ${{ github.workflow }}-${{ github.ref }}
     cancel-in-progress: true
 env:
-    # Configure default GenAIScript models
-    # using Ollama's models
-    GENAISCRIPT_DEFAULT_MODEL: ollama:qwen2.5-coder:7b
-    GENAISCRIPT_DEFAULT_SMALL_MODEL: ollama:qwen2.5-coder:1.5b
-    GENAISCRIPT_DEFAULT_VISION_MODEL: ollama:llama3.2-vision:11b
+    GENAISCRIPT_DEFAULT_REASONING_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_MODEL }}
+    GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_MODEL }}
+    GENAISCRIPT_DEFAULT_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_VISION_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_VISION_MODEL }}
 jobs:
     build:
         runs-on: ubuntu-latest
@@ -40,11 +41,11 @@ jobs:
               run: git fetch origin && git pull origin main:main
             - name: genaiscript pr-describe
               continue-on-error: true
-              run: node packages/cli/built/genaiscript.cjs run pr-describe --out ./temp/genai/pr-describe -prd --out-trace $GITHUB_STEP_SUMMARY
+              run: node packages/cli/built/genaiscript.cjs run pr-describe --out ./temp/genai/pr-describe -prd -m reasoning --out-trace $GITHUB_STEP_SUMMARY
               env:
                   GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
             - name: genaiscript pr-review
-              run: node packages/cli/built/genaiscript.cjs run pr-review --out ./temp/genai/pr-review -prc --out-trace $GITHUB_STEP_SUMMARY
+              run: node packages/cli/built/genaiscript.cjs run pr-review --out ./temp/genai/pr-review -prc -m reasoning --out-trace $GITHUB_STEP_SUMMARY
               continue-on-error: true
               env:
                   GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

diff --git a/.github/workflows/ollama.yml b/.github/workflows/ollama.yml
@@ -14,6 +14,12 @@ on:
 concurrency:
     group: ${{ github.workflow }}-${{ github.ref }}-ollama
     cancel-in-progress: true
+env:
+    GENAISCRIPT_DEFAULT_REASONING_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_MODEL }}
+    GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_REASONING_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_MODEL }}
+    GENAISCRIPT_DEFAULT_SMALL_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_SMALL_MODEL }}
+    GENAISCRIPT_DEFAULT_VISION_MODEL: ${{ vars.GENAISCRIPT_DEFAULT_VISION_MODEL }}
 jobs:
     tests:
         runs-on: ubuntu-latest

diff --git a/README.md b/README.md
@@ -208,7 +208,7 @@ const { files } = await retrieval.vectorSearch("cats", "**/*.md")
 
 ### 🐙 GitHub Models and GitHub Copilot
 
-Run models through [GitHub Models](https://microsoft.github.io/genaiscript/getting-started/configuration#github) or [GitHub Copilot](https://microsoft.github.io/genaiscript/getting-started/configuration/#github-copilot).
+Run models through [GitHub Models](https://microsoft.github.io/genaiscript/getting-started/configuration#github) or [GitHub Copilot](https://microsoft.github.io/genaiscript/getting-started/configuration/#github_copilot_chat).
 
 ```js
 script({ ..., model: "github:gpt-4o" })

diff --git a/docs/src/content/docs/getting-started/configuration.mdx b/docs/src/content/docs/getting-started/configuration.mdx
@@ -833,7 +833,7 @@ script({ model: "google:gemini-1.5-pro-latest" })
 
 <LLMProviderFeatures provider="google" />
 
-## GitHub Copilot Chat Models <a id="github-copilot" href=""></a>
+## GitHub Copilot Chat Models <a id="github_copilot_chat" href=""></a>
 
 If you have access to **GitHub Copilot Chat in Visual Studio Code**,
 GenAIScript will be able to leverage those [language models](https://code.visualstudio.com/api/extension-guides/language-model) as well.
@@ -1256,7 +1256,7 @@ script({
 
 <LLMProviderFeatures provider="deepseek" />
 
-## LMStudio
+## LM Studio <a href="" id="lmstudio" />
 
 The `lmstudio` provider connects to the [LMStudio](https://lmstudio.ai/) headless server.
 and allows to run local LLMs.
@@ -1319,7 +1319,7 @@ script({
 
 <LLMProviderFeatures provider="lmstudio" />
 
-### LMStudio and Hugging Face Models
+### LM Studio and Hugging Face Models
 
 Follow [this guide](https://huggingface.co/blog/yagilb/lms-hf) to load Hugging Face models into LMStudio.
 
@@ -1461,6 +1461,16 @@ OPENROUTER_SITE_URL=... # populates HTTP-Referer header
 OPENROUTER_SITE_NAME=... # populate X-Title header
 ```
 
+## LiteLLM
+
+The [LiteLLM](https://docs.litellm.ai/) proxy gateway provides a OpenAI compatible API for running models locally.
+Configure the `LITELLM_...` keys to set the key and optionally the base url.
+
+```txt title=".env"
+LITELLM_API_KEY="..."
+#LITELLM_API_BASE="..."
+```
+
 ## Hugging Face Transformer.js <a href="" id="transformers" />
 
 This `transformers` provider runs models on device using [Hugging Face Transformers.js](https://huggingface.co/docs/transformers.js/index).

diff --git a/docs/src/content/docs/index.mdx b/docs/src/content/docs/index.mdx
@@ -373,7 +373,7 @@ const res = await safety.detectPromptInjection(env.vars.input)
 <Card title="GitHub Models and GitHub Copilot" icon="github">
 
 Run models through GitHub using [GitHub Models](/genaiscript/getting-started/configuration#github)
-or [GitHub Copilot](/genaiscript/getting-started/configuration/#github-copilot).
+or [GitHub Copilot](/genaiscript/getting-started/configuration/#github_copilot_chat).
 
 ```js wrap
 script({ ..., model: "github:gpt-4o" })

diff --git a/docs/src/content/docs/reference/scripts/system.mdx b/docs/src/content/docs/reference/scripts/system.mdx
@@ -722,6 +722,24 @@ DIFF ./file4.ts:
 `````
 
 
+### `system.english`
+
+Use english output
+
+
+
+
+
+`````js wrap title="system.english"
+system({
+    title: "Use english output",
+})
+$`## English output
+Use English in the output of the system. Use English in the reasoning output as well.`
+
+`````
+
+
 ### `system.explanations`
 
 Explain your answers

diff --git a/packages/cli/README.md b/packages/cli/README.md
@@ -205,7 +205,7 @@ const { files } = await retrieval.vectorSearch("cats", "**/*.md")
 
 ### 🐙 GitHub Models and GitHub Copilot
 
-Run models through [GitHub Models](https://microsoft.github.io/genaiscript/getting-started/configuration#github) or [GitHub Copilot](https://microsoft.github.io/genaiscript/getting-started/configuration/#github-copilot-in-visual-studio-code).
+Run models through [GitHub Models](https://microsoft.github.io/genaiscript/getting-started/configuration#github) or [GitHub Copilot](https://microsoft.github.io/genaiscript/getting-started/configuration/#github_copilot_chat).
 
 ```js
 script({ ..., model: "github:gpt-4o" })

diff --git a/packages/cli/src/run.ts b/packages/cli/src/run.ts
@@ -195,7 +195,7 @@ export async function runScriptInternal(
     const removeOut = options.removeOut
     const cancellationToken = options.cancellationToken
     const jsSource = options.jsSource
-    const fallbackTools = !!options.fallbackTools
+    const fallbackTools = options.fallbackTools
     const logprobs = options.logprobs
     const topLogprobs = normalizeInt(options.topLogprobs)
     const fenceFormat = options.fenceFormat

diff --git a/packages/cli/src/server.ts b/packages/cli/src/server.ts
@@ -189,7 +189,9 @@ export async function startServer(options: {
             providers: await resolveLanguageModelConfigurations(undefined, {
                 token: false,
                 error: true,
+                models: true,
             }),
+            modelAliases: runtimeHost.modelAliases,
             remote: remote
                 ? {
                       url: remote,

diff --git a/packages/core/src/config.ts b/packages/core/src/config.ts
@@ -18,6 +18,7 @@ import { deleteEmptyValues } from "./cleaners"
 import { errorMessage } from "./error"
 import schema from "../../../docs/public/schemas/config.json"
 import defaultConfig from "./config.json"
+import { CancellationOptions } from "./cancellation"
 
 export async function resolveGlobalConfiguration(
     dotEnvPath?: string
@@ -100,7 +101,11 @@ export async function resolveGlobalConfiguration(
  */
 export async function resolveLanguageModelConfigurations(
     provider: string,
-    options?: { token?: boolean; error?: boolean; models?: boolean }
+    options?: {
+        token?: boolean
+        error?: boolean
+        models?: boolean
+    } & CancellationOptions
 ): Promise<ResolvedLanguageModelConfiguration[]> {
     const { token, error, models } = options || {}
     const res: ResolvedLanguageModelConfiguration[] = []
@@ -118,19 +123,19 @@ export async function resolveLanguageModelConfigurations(
             if (conn) {
                 // Mask the token if the option is set
                 let listError = ""
-                if (!token && conn.token) conn.token = "***"
                 if (models) {
                     const lm = await resolveLanguageModel(modelProvider.id)
                     if (lm.listModels) {
-                        const models = await lm.listModels(conn, {})
+                        const models = await lm.listModels(conn, options)
                         if (models.ok) conn.models = models.models
                         else
                             listError =
-                                errorMessage(listError) ||
-                                "failed to llist models"
+                                errorMessage(models.error) ||
+                                "failed to list models"
                     }
                 }
-                if (!listError || error)
+                if (!token && conn.token) conn.token = "***"
+                if (!listError || error || provider)
                     res.push(
                         deleteEmptyValues({
                             provider: conn.provider,
@@ -143,7 +148,7 @@ export async function resolveLanguageModelConfigurations(
                     )
             }
         } catch (e) {
-            if (error)
+            if (error || provider)
                 // Capture and store any errors encountered
                 res.push({
                     provider: modelProvider.id,

diff --git a/packages/core/src/expander.ts b/packages/core/src/expander.ts
@@ -21,7 +21,7 @@ import { addToolDefinitionsMessage, appendSystemMessage } from "./chat"
 import { importPrompt } from "./importprompt"
 import { parseModelIdentifier } from "./models"
 import { runtimeHost } from "./host"
-import { resolveSystems } from "./systems"
+import { addFallbackToolSystems, resolveSystems } from "./systems"
 import { GenerationOptions } from "./generation"
 import {
     AICIRequest,
@@ -356,7 +356,7 @@ export async function expandTemplate(
         trace.endDetails()
     }
 
-    if (systems.includes("system.tool_calls")) {
+    if (addFallbackToolSystems(systems, tools, template, options)) {
         addToolDefinitionsMessage(messages, tools)
         options.fallbackTools = true
     }

diff --git a/packages/core/src/genaisrc/system.english.genai.js b/packages/core/src/genaisrc/system.english.genai.js
@@ -0,0 +1,5 @@
+system({
+    title: "Use english output",
+})
+$`## English output
+Use English in the output of the system. Use English in the reasoning output as well.`
diff --git a/packages/core/src/llms.json b/packages/core/src/llms.json
@@ -7,6 +7,7 @@
             "bearerToken": true,
             "transcribe": true,
             "speech": true,
+            "listModels": true,
             "aliases": {
                 "large": "gpt-4o",
                 "small": "gpt-4o-mini",

diff --git a/packages/core/src/openai.ts b/packages/core/src/openai.ts
@@ -456,7 +456,7 @@ export const OpenAIListModels: ListModelsFunction = async (cfg, options) => {
             return {
                 ok: false,
                 status: res.status,
-                error: serializeError(res.statusText),
+                error: serializeError(await res.json()),
             }
         const { data } = (await res.json()) as {
             object: "list"