Merge staging into prod — 26.04 release

.github/workflows/deploy-docs-prod.yaml

@@ -117,7 +117,7 @@ jobs:
         run: aws configure set aws_access_key_id ${{ secrets.AWS_ACCESS_KEY_ID_PROD }} && aws configure set aws_secret_access_key ${{ secrets.AWS_SECRET_ACCESS_KEY_PROD }} && aws configure set default.region us-east-1
 
       - name: Deploy docs prod site
-        run: aws s3 sync site/_site s3://validmind-docs-prod/site --delete --exclude "installation/omnibus/*" --exclude "installation/helm-repo/*" --exclude "llm/*" && aws cloudfront create-invalidation --distribution-id E2BGG3USKQTR9W --paths "/*" --no-cli-pager
+        run: aws s3 sync site/_site s3://validmind-docs-prod/site --delete --exclude "installation/omnibus/*" --exclude "installation/helm-repo/*" --exclude "notebooks/EXECUTED/*" --exclude "llm/*" && aws cloudfront create-invalidation --distribution-id E2BGG3USKQTR9W --paths "/*" --no-cli-pager
 
       # Release headroom and shrink before final lightweight steps & post-job
       - name: Release reserve & shrink

.github/workflows/deploy-docs-staging.yaml

@@ -120,7 +120,7 @@ jobs:
         run: aws configure set aws_access_key_id ${{ secrets.AWS_ACCESS_KEY_ID_STAGING }} && aws configure set aws_secret_access_key ${{ secrets.AWS_SECRET_ACCESS_KEY_STAGING }} && aws configure set default.region us-west-2
 
       - name: Deploy docs staging site
-        run: aws s3 sync site/_site s3://validmind-docs-staging/site --delete --exclude "installation/helm-repo/*" --exclude "pr_previews/*" --exclude "llm/*" && aws cloudfront create-invalidation --distribution-id ESWVTZYFL873V --paths "/*" --no-cli-pager
+        run: aws s3 sync site/_site s3://validmind-docs-staging/site --delete --exclude "installation/helm-repo/*" --exclude "pr_previews/*" --exclude "notebooks/EXECUTED/*" --exclude "llm/*" && aws cloudfront create-invalidation --distribution-id ESWVTZYFL873V --paths "/*" --no-cli-pager
 
       # Release headroom and shrink before final lightweight steps & post-job
       - name: Release reserve & shrink

.github/workflows/validate-docs-site.yaml

@@ -130,7 +130,7 @@ jobs:
       run: aws configure set aws_access_key_id ${{ secrets.AWS_ACCESS_KEY_ID_STAGING }} && aws configure set aws_secret_access_key ${{ secrets.AWS_SECRET_ACCESS_KEY_STAGING }} && aws configure set default.region us-east-1
 
     - name: Deploy PR preview
-      run: aws s3 sync site/_site s3://validmind-docs-staging/site/pr_previews/${{ github.head_ref }} --delete && aws cloudfront create-invalidation --distribution-id ESWVTZYFL873V --paths "/*" --no-cli-pager
+      run: aws s3 sync site/_site s3://validmind-docs-staging/site/pr_previews/${{ github.head_ref }} --delete --exclude "notebooks/EXECUTED/*" && aws cloudfront create-invalidation --distribution-id ESWVTZYFL873V --paths "/*" --no-cli-pager
 
     - name: Post comment with preview URL
       uses: actions/github-script@v6

.gitignore

@@ -36,9 +36,6 @@ site/validmind-docs.yaml
 # Python API docs are now generated on the fly
 site/validmind
 
-# Generated template schema documentation
-site/guide/templates/_template-schema-generated.qmd
-
 # Cursor rules
 .cursor/rules/
 .cursor/skills/

README.md

@@ -150,7 +150,7 @@ Some documentation content is auto-generated from backend source files. These sc
 
 #### Template schema documentation
 
-The template schema reference in `site/guide/templates/customize-document-templates.qmd` is auto-generated from the backend JSON Schema. CI workflows generate this automatically, but you can also regenerate locally:
+The template schema reference in `site/guide/templates/customize-document-templates.qmd` is auto-generated from the backend JSON Schema, overwriting any baseline output checked into this repo. You can also regenerate locally and commit:
 
 ```bash
 cd site
@@ -172,7 +172,7 @@ make template-schema-docs
 The script reads from:
 - `backend/src/backend/templates/documentation/model_documentation/mdd_template_schema_v5.json` — template schema definition
 
-Output: `site/guide/templates/_template-schema-generated.qmd`
+Output: Content is injected directly into `site/guide/templates/customize-document-templates.qmd` between marker comments.
 
 #### Stylesheet organization (IN PROGRESS)
 

site/releases/breaking-changes/breaking-changes.qmd → internal/releases-archive/breaking-changes/breaking-changes.qmd

@@ -1,4 +1,7 @@
 ---
+# Copyright © 2023-2026 ValidMind Inc. All rights reserved.
+# Refer to the LICENSE file in the root of this repository for details.
+# SPDX-License-Identifier: AGPL-3.0 AND ValidMind Commercial
 title: "Breaking changes and deprecations"
 date: last-modified
 categories: ["breaking changes", "deprecations"]

scripts/generate_template_schema_docs.py

@@ -30,7 +30,10 @@
 BACKEND_ROOT = Path(os.environ.get("BACKEND_ROOT", REPO_ROOT.parent / "backend"))
 
 SCHEMA_FILE = BACKEND_ROOT / "src/backend/templates/documentation/model_documentation/mdd_template_schema_v5_ui.json"
-OUTPUT_FILE = REPO_ROOT / "site/guide/templates/_template-schema-generated.qmd"
+TARGET_FILE = REPO_ROOT / "site/guide/templates/_template-schema-generated.qmd"
+
+# Minimum expected file size in bytes (sanity check for valid output)
+MIN_OUTPUT_SIZE = 40000
 
 # CSS stylesheets to link in the generated HTML
 STYLESHEETS = [
@@ -62,7 +65,7 @@ def main():
     print(f"Generating schema documentation from {SCHEMA_FILE}")
 
     # Create temporary output
-    temp_output = OUTPUT_FILE.with_suffix(".tmp.html")
+    temp_output = TARGET_FILE.with_suffix(".tmp.html")
 
     subprocess.run([
         "generate-schema-doc",
@@ -98,6 +101,20 @@ def main():
         '',
         html_content
     )
+    # Strip the title tag (not needed when embedded in Quarto page)
+    html_content = re.sub(
+        r'\s*<title>[^<]*</title>\s*',
+        '\n',
+        html_content
+    )
+    # Strip h1 headings (the page already has a heading from Quarto)
+    html_content = re.sub(
+        r'<h1>[^<]*</h1>',
+        '',
+        html_content
+    )
+    # Clean up multiple consecutive blank lines
+    html_content = re.sub(r'\n{3,}', '\n\n', html_content)
 
     # Build stylesheet link tags
     stylesheet_links = '\n'.join(
@@ -124,34 +141,45 @@ def main():
         '</div></body>'
     )
 
-    # Copyright header to place before the raw HTML block
-    copyright_header = """<!-- Copyright © 2023-2026 ValidMind Inc. All rights reserved.
+    # Strip leading/trailing whitespace from HTML content
+    html_content = html_content.strip()
+
+    # Build the output content (raw HTML block for Quarto include)
+    output_content = f"""<!-- Copyright © 2023-2026 ValidMind Inc. All rights reserved.
 Refer to the LICENSE file in the root of this repository for details.
 SPDX-License-Identifier: AGPL-3.0 AND ValidMind Commercial
 
 This file is auto-generated by scripts/generate_template_schema_docs.py
 Do not edit directly. Re-run the script to update.
 
-Source: backend/src/backend/templates/documentation/model_documentation/mdd_template_schema_v5.json
--->"""
-
-    # Wrap HTML in Quarto raw HTML block for .qmd file, with comment before
-    qmd_content = f"""{copyright_header}
+Source: {SCHEMA_FILE.relative_to(BACKEND_ROOT.parent)}
+-->
 
 ```{{=html}}
 {html_content}
 ```
 """
 
-    # Write final output
-    OUTPUT_FILE.parent.mkdir(parents=True, exist_ok=True)
-    with open(OUTPUT_FILE, "w") as f:
-        f.write(qmd_content)
+    # Validate output before writing
+    if len(output_content) < MIN_OUTPUT_SIZE:
+        print(f"Error: Generated output is too small ({len(output_content)} bytes)")
+        print("This likely indicates a generation failure.")
+        temp_output.unlink(missing_ok=True)
+        sys.exit(1)
+
+    if "<html" not in html_content or "</html>" not in html_content:
+        print("Error: Generated output does not contain valid HTML structure")
+        temp_output.unlink(missing_ok=True)
+        sys.exit(1)
+
+    # Write to target file
+    TARGET_FILE.write_text(output_content)
 
     # Clean up temp file
     temp_output.unlink()
 
-    print(f"Generated {OUTPUT_FILE}")
+    print(f"Generated template schema documentation: {TARGET_FILE}")
+    print(f"Output size: {len(output_content)} bytes")
 
 
 if __name__ == "__main__":

site/Makefile

@@ -427,7 +427,7 @@ release-notes:
 template-schema-docs:
 	@echo "\nGenerating template schema documentation ..."
 	@if [ ! -d "$(SRC_ROOT)/backend" ]; then echo "Error: Backend not cloned. Run 'make clone' first."; exit 1; fi
-	@pip install -q json-schema-for-humans
+	@python -m pip install -q json-schema-for-humans
 	@BACKEND_ROOT=$(SRC_ROOT)/backend python ../scripts/generate_template_schema_docs.py
 
 test-descriptions:

site/_extensions/category-filter/_extension.yml

@@ -0,0 +1,6 @@
+title: Category Filter
+author: Nik Richers
+version: 1.0.0
+contributes:
+  filters:
+    - category-filter.lua

site/_extensions/category-filter/category-filter.lua

@@ -0,0 +1,59 @@
+-- Category Filter Lua Extension
+-- Reads listing-filter from YAML frontmatter and injects config as a data attribute
+
+local function meta_to_native(val)
+  if val == nil then
+    return nil
+  end
+
+  local mt = getmetatable(val)
+
+  -- Check for pandoc List type (used for YAML arrays)
+  if mt and mt.__name == "List" then
+    local arr = {}
+    for i, item in ipairs(val) do
+      arr[i] = meta_to_native(item)
+    end
+    return arr
+  end
+
+  -- Check for pandoc Inlines (YAML string values become Inlines)
+  if mt and (mt.__name == "Inlines" or mt.__name == "Blocks") then
+    return pandoc.utils.stringify(val)
+  end
+
+  -- Plain Lua types
+  local t = type(val)
+  if t == "string" or t == "number" or t == "boolean" then
+    return val
+  elseif t == "table" then
+    -- Check if it's an array-like table
+    if #val > 0 and val[1] ~= nil then
+      local arr = {}
+      for i, item in ipairs(val) do
+        arr[i] = meta_to_native(item)
+      end
+      return arr
+    else
+      -- It's a map/object
+      local obj = {}
+      for k, v in pairs(val) do
+        obj[k] = meta_to_native(v)
+      end
+      return obj
+    end
+  end
+
+  -- Fallback: try stringify
+  return pandoc.utils.stringify(val)
+end
+
+function Meta(meta)
+  if meta['listing-filter'] then
+    local config = meta_to_native(meta['listing-filter'])
+    local json = quarto.json.encode(config)
+    local escaped = json:gsub("'", "'")
+    quarto.doc.include_text("after-body",
+      '<div id="listing-filter-config" style="display:none" data-config=\'' .. escaped .. '\'></div>')
+  end
+end

site/_quarto.yml

@@ -4,6 +4,10 @@
 
 project:
   type: website
+  render:
+    - "**/*.qmd"
+    - "**/*.ipynb"
+    - "**/*.md"
 
 metadata-files:
   # - _drafts.yaml

site/about/glossary/mcp/_mcp-server.qmd

@@ -0,0 +1,6 @@
+<!-- Copyright © 2023-2026 ValidMind Inc. All rights reserved.
+Refer to the LICENSE file in the root of this repository for details.
+SPDX-License-Identifier: AGPL-3.0 AND ValidMind Commercial -->
+
+MCP server
+: A service that exposes tools and resources to AI assistants via the MCP protocol. The {{< var vm.product >}} MCP server provides access to model inventory operations.

site/about/glossary/mcp/_mcp-tool.qmd

@@ -0,0 +1,6 @@
+<!-- Copyright © 2023-2026 ValidMind Inc. All rights reserved.
+Refer to the LICENSE file in the root of this repository for details.
+SPDX-License-Identifier: AGPL-3.0 AND ValidMind Commercial -->
+
+MCP tool
+: An operation exposed by an MCP server that AI assistants can invoke to perform actions, such as querying models or updating artifacts.

site/about/glossary/mcp/_mcp.qmd

@@ -0,0 +1,6 @@
+<!-- Copyright © 2023-2026 ValidMind Inc. All rights reserved.
+Refer to the LICENSE file in the root of this repository for details.
+SPDX-License-Identifier: AGPL-3.0 AND ValidMind Commercial -->
+
+MCP (Model Context Protocol)
+: An open standard that enables AI assistants to connect to external data sources and tools through a unified protocol, allowing natural language interactions with structured systems.

site/developer/model-documentation/install-and-initialize-validmind-for-r.qmd

@@ -12,12 +12,12 @@ listing:
     fields: [title, description]
     grid-columns: 2
     contents:
-    - path: https://github.com/validmind/validmind-library/blob/main/notebooks/quickstart/quickstart_model_documentation.Rmd
-      title: "Quickstart for model documentation"
-      description: "End-to-end model documentation workflow in R: load data, preprocess, train a GLM model, and run the full documentation test suite."
-    - path: https://github.com/validmind/validmind-library/blob/main/notebooks/quickstart/quickstart_model_validation.Rmd
-      title: "Quickstart for model validation"
-      description: "End-to-end model validation workflow in R: load data, run data quality tests, train a champion GLM model, and run model evaluation tests."
+      - path: https://github.com/validmind/validmind-library/tree/main/notebooks/quickstart/
+        title: "Quickstart for model documentation"
+        description: "End-to-end model documentation workflow in R: load data, preprocess, train a GLM model, and run the full documentation test suite."
+      - path: https://github.com/validmind/validmind-library/tree/main/notebooks/quickstart/
+        title: "Quickstart for model validation"
+        description: "End-to-end model validation workflow in R: load data, run data quality tests, train a champion GLM model, and run model evaluation tests."
 ---
 
 Use the ValidMind R package to document and validate models built in R. The package interfaces with the {{< var validmind.developer >}} through `reticulate`, giving you access to the full Python API from R.

site/developer/supported-models-and-frameworks.qmd

@@ -81,6 +81,12 @@ Retrieval-augmented generation pipelines.
 
 - Tests: RAGAS integration
 - Refer to [RAG evaluation](#rag-evaluation) for dataset requirements and available tests
+
+## Agentic AI
+LLM-based agents that use tools and multi-step reasoning.
+
+- Scorers: DeepEval integration
+- Refer to [Agentic AI evaluation](#agentic-ai-evaluation) for available scorers
 :::
 
 ## Machine learning models
@@ -465,6 +471,28 @@ run_test(
 - **Context Precision** — Measures relevance of retrieved contexts
 - **Answer Relevancy** — Assesses if the response addresses the query
 
+## Agentic AI evaluation
+
+For agentic AI systems, the {{< var validmind.developer >}} integrates with [DeepEval](https://docs.confident-ai.com/) for trace-based evaluation of tool usage and reasoning.
+
+### Available agentic scorers
+
+- **TaskCompletion** — Assesses whether the agent achieves the requested outcome
+- **PlanQuality** — Measures if generated plans are logical, complete, and efficient
+- **PlanAdherence** — Evaluates whether the agent follows its plan during execution
+- **ToolCorrectness** — Validates that the agent invokes the expected tools
+- **ArgumentCorrectness** — Checks if arguments passed to tools are correct
+
+### Scorer requirements
+
+| Scorer | Requires model | Parameters |
+|--------|----------------|------------|
+| TaskCompletion | Yes (`predict_fn`) | `input_column` |
+| PlanQuality | Yes (`predict_fn`) | `input_column` |
+| PlanAdherence | Yes (`predict_fn`) | `input_column` |
+| ToolCorrectness | No | `input_column`, `expected_tools_called_column`, `actual_tools_called_column` |
+| ArgumentCorrectness | No | `input_column`, `actual_tools_called_column` |
+
 ## Python and dependency compatibility
 
 The {{< var validmind.developer >}} requires:

site/guide/_sidebar.yaml

@@ -42,9 +42,10 @@ website:
           contents:
           - guide/integrations/manage-secrets.qmd
           - guide/integrations/configure-connections.qmd
-          - guide/integrations/configure-analytics-exports.qmd
           - guide/integrations/implement-custom-integrations.qmd
           - guide/integrations/link-external-models.qmd
+          - guide/mcp/connect-ai-assistants-via-mcp.qmd
+          - guide/integrations/configure-analytics-exports.qmd
           - guide/integrations/integrations-examples.qmd
         - text: "---"
         - text: "Workflows"

site/guide/integrations/managing-integrations.qmd

@@ -33,13 +33,15 @@ graph TD
     E[Public REST API]
     F[Testing & documentation]
     G[Service integration]
+    H[AI assistants]
 
     A <--> B
     A <--> C
     A <--> D
     A <--> E
     A <--> F
     A <--> G
+    A <--> H
 ```
 
 <!-- The {{< var validmind.platform >}} supports various ways to integrate with external systems and services to make changes in one system visible in the other, keeping information synchronized: -->
@@ -194,6 +196,29 @@ Link to tickets in external systems and keep them synchronized. Attach tickets t
 
 ::::
 
+---
+
+### AI assistants
+
+Connect AI assistants to {{< var vm.product >}} using the Model Context Protocol (MCP). Query and manage your model inventory through natural language conversations.[^7]
+
+**Supported**
+
+:::: {.flex .flex-wrap .justify-around}
+
+::: {.w-33-ns}
+- Claude Desktop
+:::
+
+::: {.w-33-ns}
+- Cursor IDE
+:::
+
+::: {.w-33-ns}
+&nbsp;
+:::
+
+::::
 
 ## Key concepts
 
@@ -233,4 +258,5 @@ Link to tickets in external systems and keep them synchronized. Attach tickets t
 [^3]: [Introduction to workflows](/guide/workflows/introduction-to-workflows.qmd)
 [^4]: [Public REST API](/reference/validmind-rest-api-vm.qmd)
 [^5]: [Testing & documentation](/developer/validmind-library.qmd)
-[^6]: [Configure analytics exports](/guide/integrations/configure-analytics-exports.qmd)
+[^6]: [Configure analytics exports](/guide/integrations/configure-analytics-exports.qmd)
+[^7]: [Connect AI assistants](/guide/mcp/connect-ai-assistants-via-mcp.qmd)