Wren AI 0.29.2 adds Databricks while hardening a legacy application line

A connector release with an architectural footnote

Wren AI published version 0.29.2 on August 5 with Databricks added to its list of supported databases. The release also changes the AI service’s knowledge retrieval path to use Ibis and adds system-prompt handling to its LLM provider and pipelines. Those are material additions for a natural-language analytics application: the first expands the warehouses it can query, while the latter two affect how context reaches query-generation workflows. Wren AI’s release notes list the changes.

The version number alone is easy to misread, however. Release 0.29.2 targets the repository’s legacy/v1 branch. Wren AI’s repository-level changelog says the former GenBI application—including its AI service, UI, launcher and deployment assets—was archived there during a 2026 second-quarter consolidation. The main branch now houses the imported Wren engine packages and a new multi-license repository layout. The repository changelog documents that split.

Security maintenance dominates the change list

Most entries in 0.29.2 are dependency upgrades and vulnerability fixes rather than new NL2SQL behavior. The release notes identify fixes for several named CVEs in UI and service dependencies, including fast-xml-parser, lodash, grpc-go and LiteLLM, alongside upgrades to Next.js, aiohttp, urllib3, filelock, tar and axios. One service change upgrades sqlparse to address hangs on long tuple lists.

That maintenance matters because text-to-SQL products combine exposed web interfaces, database credentials and model gateways. A connector expansion without corresponding dependency hygiene would enlarge the operational surface. Wren AI’s release bundles both, although the notes do not provide a product-specific exploit assessment or claim that these vulnerabilities were exploited in Wren deployments.

The release also ships launcher archives for Linux, macOS and Windows. Its GitHub comparison records 39 commits between 0.29.1 and 0.29.2, accumulated over roughly eight months. That cadence helps explain the breadth of the security sweep: this is a consolidated maintenance release rather than a narrow patch.

Two Wren release streams now coexist

Wren AI has continued publishing packages from the consolidated mainline separately. On August 11 it released wren-core-py 0.7.4, declaring the engine version against which its Python bindings were built. On August 10 it released semantic-core 0.3.1 with per-invocation state for RLAC cycle detection. The repository’s releases page shows those package releases alongside the legacy application tag.

For operators, the distinction is the main takeaway. Version 0.29.2 is the relevant update for deployments still running the older self-contained Wren AI application: it adds Databricks and closes a substantial dependency backlog. It should not be interpreted as the current architecture of the consolidated Wren repository, whose core packages now advance on their own version tracks.