Brief: The web UI (ouro-webui/) is decoupled from the core ouro/ skill and is not part of the distributed package.

Context

Ourobor OS's primary interface is plain Markdown files — readable in any editor, any LLM context window, and any terminal. A web UI for browsing the wiki was prototyped in ouro-webui/ using mistune and jinja2.

The question arose whether builder.py and its templates should be bundled into the distributed ouro/ skill package so users get a dashboard out of the box.

Decision

The web UI lives in ouro-webui/ and is excluded from the distributed package. Users who want a rendered dashboard use the web UI separately. The core skill has no dependency on mistune, jinja2, or the templates.

Alternatives Considered

• Bundle web UI in ouro/: Adds mistune, jinja2, and template files to every install. Most users running the skill in an LLM coding session don't need a browser dashboard — this adds weight with no benefit for the majority use case.
• Make web UI a separate pip package: Premature. The web UI has known bugs (template inheritance broken, no navigation sidebar) and is not ready for independent distribution.
• Build web UI with JavaScript (client-side): Would eliminate Python dependencies for rendering, but adds a JS build step and larger file footprint. Over-engineered for the current stage.

Trade-offs

• Lost: Out-of-the-box visual dashboard for new users. The install experience is text-only.
• Gained: Core skill stays lightweight — pure Python stdlib plus Markdown files. The web UI can evolve (and be fixed) independently without affecting skill compatibility or version constraints.

Rationale

The wiki's primary consumer is an LLM agent reading files in a coding session, not a human browsing in a browser. The web UI is a nice-to-have for human navigation and is not on the critical path for the skill's core value proposition.