Blume 2.0 refactors documentation configuration into modular adapters, introduces agent discovery settings, and adds automated upgrade tooling.
What the product does and how it is positioned
Blume 2.0 is a major release of the documentation framework that restructures site configuration while preserving existing Markdown and MDX content. The update replaces legacy string options and nested configuration blocks with typed adapters imported from specialized subpaths covering deployment hosts, search engines, content sources, analytics tools, API references, and AI assistant backends.
The release also modernizes discoverability by organizing machine-readable AI metadata, Model Context Protocol configurations, and skills into a dedicated agents namespace. Projects can automate their migration using an interactive command-line upgrade command compatible with automated coding agents like Codex and Claude Code.
Source-supported ways to use the product
Development teams can upgrade existing documentation repositories by executing the upgrade CLI to identify obsolete settings, adjust component overrides, and update configuration schemas.
Organizations can integrate documentation from local repositories, Notion workspaces, Sanity, GitHub Releases, and API schemas into a unified, server-rendered or static documentation site.
The documented workflow, where available
Run the upgrade CLI from the project root to bump the package version and scan configuration files for necessary changes.
Convert search, deployment, content sources, API reference, analytics, and assistant blocks in blume.config.ts into imported adapter function calls.
Move inline functions and internal declarations in components.ts into separate files, importing them as discrete component references.
Execute the doctor and build CLI commands to confirm that all configuration settings and component bindings pass validation without errors.
Blume 2.0 shifts its architecture toward modular adapters across all functional integration points. Rather than specifying provider strings and credential blocks inside blume.config.ts, developers import dedicated adapter functions from specialized subpaths such as blume/search, blume/deploy, blume/sources, blume/reference, blume/analytics, and blume/ai. This architecture allows host-specific outputs for platforms like Vercel, Netlify, Cloudflare, and Node.js to be declared explicitly, removing implicit runtime server detection and obsolete command-line build flags.
To streamline adoption, Blume 2.0 includes an automated upgrade workflow via the CLI. Running the upgrade utility inspects blume.config.ts, components.ts, and package scripts, outputting exact file locations and replacements for outdated configurations. The tool can also invoke coding agents like Codex or Claude Code to resolve migration findings interactively before verifying the site with built-in doctor and build diagnostics.
Checks to run with your own material and workflow
What was checked and when
Answers based on the source-checked product record
Markdown and MDX pages require no edits during the upgrade, as Blume 2.0 changes configuration syntax rather than page content structure.
The search.boost frontmatter field, which was ignored in Blume 1, is now active and directly multiplies the search relevance ranking for the page.
Running the upgrade command with the codex or claude flag opens an interactive agent session to apply required configuration fixes automatically.
All component entries must now be imported components, file paths, or specific component descriptor objects, with inline functions and internal declarations disallowed.
Individual reference blocks are replaced by a unified reference array populated by adapter functions for OpenAPI, AsyncAPI, GraphQL, and Scalar.
