Document360
Document360 rebuilt its API for agents; now it's turning the AI inward on authoring.
◆Recent moves
- 2d ago
Publish dialog folds in AI suggestions and pre-publish checks
The Publish dialog becomes the single place an article is validated and shipped: metadata, publishing options, one-click Auto Fix on AI suggestions, and upfront warnings for broken links, unresolved comments and reader acknowledgements. Offline and Web Help exports were also brought closer to the live site, restoring heading hierarchy, breadcrumbs, table of contents and RTL direction. A consolidation release after three months of API work.
View source ↗ - 9d ago
API v3 rebuild adds OAuth 2.0 and scoped API keys
⚡ SPARKThe pivot the whole agent-readability arc was heading toward: scoped keys and OAuth turn programmatic access into something an enterprise security team can approve. Pairing it in the same release with reader-group restrictions on Eddy AI features shows the consistent instinct — widen the access surface, then fence it.
View source ↗ - 16d ago
Interactive API reference with a rebuilt Try It! console
The API reference became a working console — three-pane layout, live schema validation as you type, in-console authorization, reusable session variables, and Eddy AI answering endpoint questions inline. It set up the v3 release two weeks later; Document360 is building its documentation product and its API as one thing.
View source ↗ - 1mo ago
Find and replace spans variables, snippets, and links
Bulk content operations mature: a single find-and-replace pass now reaches variables, snippets, glossary definitions and hyperlink URLs, with workspace and category filters. The quieter item is readers being able to open an article directly in ChatGPT or Claude — an early move in the same agent-readability thread as llms.txt and MCP.
View source ↗ - 1mo ago
Native Mermaid diagrams and automatic llms.txt generation
Mermaid support brings diagramming in-editor with the syntax preserved so diagrams stay editable, while automatic llms.txt generation makes the knowledge base discoverable to agents with no author effort at all. Two features for different audiences, both removing an external dependency.
View source ↗ - 2mo ago
MCP server gains publishing and workflow controls
MCP moved past read-and-draft into the full content lifecycle — list workflow statuses, assign reviewers, move articles through stages, publish and unpublish, all from an assistant conversation. This is the release that made the MCP server somewhere work gets completed rather than just a search endpoint.
View source ↗