ChatWalaʻau v0.100.0

This release adds two new features to ChatWalaʻau: the Microsoft Foundry model provider and Ontology Concept Modeling.

Highlights

Microsoft Foundry Model Provider (v0.100.0)

• Serve models from a Microsoft Foundry project. List your Foundry model deployment names in FOUNDRY_MODELS and point FOUNDRY_PROJECT_ENDPOINT at your project endpoint (https://<resource>.services.ai.azure.com/api/projects/<project>). The models appear in the model selector alongside your Azure OpenAI, Anthropic, and OpenAI models -- OpenAI deployments (gpt-5.x / o-series) as well as other Foundry catalog families such as DeepSeek.
• Sign in with Entra ID -- no new credentials. Foundry reuses the same Azure sign-in you already use for Azure OpenAI: az login on your workstation, a managed identity on Azure-hosted compute, or the default credential chain (AZURE_CREDENTIAL_MODE, AZURE_TENANT_ID). There is no API-key option -- Foundry project endpoints authenticate with Entra ID; grant your identity a Foundry data-plane role (for example Azure AI User) on the project.
• Web search and structured output included. Foundry models can use the hosted web search tool (with citations) and structured JSON output, just like the other providers.
• Controls that match the model. OpenAI reasoning deployments expose the same per-message reasoning-effort and verbosity controls as Azure OpenAI reasoning models; other families (for example DeepSeek) simply show no generation controls, and requests never carry parameters those models reject.
• Opt-in and fully backward compatible. The provider is inert until FOUNDRY_MODELS is set; with it unset, the model list, default model, and all behavior are unchanged, and no operator action is required to upgrade.
• Note. This native Foundry provider is separate from the existing Claude-on-Foundry option (ANTHROPIC_FOUNDRY_*). Background responses are not yet enabled for Foundry models in this release.

Ontology Concept Modeling (v0.99.0)

• A new Ontology workspace. A sidebar icon (next to Declarative Agents, shown when the feature is enabled) opens a full-screen workspace with three resizable panes: your ontology catalog on the left, a visual design canvas in the center, and a Detail / Search pane on the right.
• Design on a node canvas. Entities are circles (emoji or initial inside, label below) with connectors on all four sides, so relationships attach from any direction; edges are directional with a cardinality (1:1, 1:N, N:1, N:M), and typed properties can be marked as key attributes (a key badge appears on the node).
• Neighborhood selection. Clicking an entity also lights up its incoming/outgoing relationships and the connected entities; clicking a relationship lights up its endpoints and their edges. Click an empty area to clear.
• Search by SPARQL or plain language. Write SPARQL in a syntax-highlighted editor (a Reset button restores the default query), or just type a question -- the send button converts it to SPARQL for you and puts the generated query in the editor so you can refine it. Table results highlight the matching entities on the canvas in strong color while everything else dims.
• Standard RDF in and out, with automatic backups. Import Turtle (.ttl) and RDF/XML (.rdf/.owl); export clean Turtle. Every save writes a timestamped backup first, and deleting an ontology asks for confirmation and keeps a backup on disk.
• Ask the assistant about your ontologies. In any chat, the assistant can list your ontologies, pick the right one from your catalog descriptions, query it in natural language, and answer with the matching RDF -- clearly fenced and size-capped so it never floods the conversation. Everything on the query side is read-only by design.
• Opt-in and fully backward compatible. The whole feature is off by default; enable it with ONTOLOGY_ENABLED=true (optional ONTOLOGY_DIR=.ontologies). With the flag unset, behavior is unchanged and no operator action is required to upgrade.

Included in this release

This release includes implementation work from:

• v0.99.0 through v0.100.0

Notes

• Internal design management content and architecture control artifacts are intentionally omitted from this release note.

Installation

See the repository README for setup and usage instructions.

Version

• Release version: 0.100.0
• Previous release: 0.98.1