DocGen - AI Documentation Generator

Generate Documentation

Input a JIRA ticket key. The AI engine fetches the full hierarchy and produces comprehensive functional documentation.

AI Provider:

Provider capabilities

Cursor CLI runs through the bundled Cursor agent in the worker. Tool use during generation depends on the configured generator; final AI review may attach read-only JIRA tools when the runtime supports them.
Anthropic Claude enables agentic tool calls during child-section generation (JIRA search, issue details, parent, issuelinks). When Confluence credentials are configured in the backend, Confluence search and page read tools are also available in that loop.

Publish to Confluence

Use images and Figma designs

Post Generation AI Review

When enabled, the final document goes through an additional AI review pass that cleans up formatting, removes empty placeholder sections, and ensures consistent heading structure. Disabling this saves one large AI call and speeds up generation.

Strict grounding

When ON, the AI is told to OMIT speculative subsections (Filtering / Mobile / Accessibility / Analytics) when no ticket evidence exists, and to write a visible "not enough information" note for tangible components instead of inventing. The verifier escalates SYNTHESIZED findings to HALLUCINATION.

Browse linked JIRA tickets

When ON, the orchestrator pulls depth-1 related tickets (issuelinks, parent, siblings) before generating, capped at 20 per generation. Renders them in a clearly-labelled "RELATED CONTEXT" section so the AI uses them as enrichment, not as new spec.

Incremental Regeneration

Compares JIRA content against the last generation and only re-processes sections that have changed, significantly reducing generation time and cost. Only effective when a previous version exists for this ticket.

Generation History

0 records

FromTo