nebari-dev / nebari-dev/nebari-frames

Import Frame: turn unstructured context into a structured Frame with an LLM

Open
#52 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

needs: discussion 💬 type: enhancement 💅🏼
Dominant language
Go
Stars
2
Forks
1
Avg merge
10h 55m
Merged PRs (30d)
9

Description

Motivation

Two pieces of feedback point the same direction:

  • Authoring a Frame through the full typed form is overwhelming.
  • Orgs already have the context written up elsewhere (e.g. Google Docs) in prose that isn't in the Frame format.

We want a low-friction path: bring existing prose in, let an LLM map it into the Frame schema, review it, and create the Frame. This lowers the barrier for non-technical authors and makes it easy to migrate context that already exists.

Proposal (surface - do one or both; web is the priority)

  • Web import flow (priority): in the web app, paste or upload the source text. A "draft with AI" step produces structured slots, which pre-fill the authoring form for the user to review and adjust before publishing. This targets the "form is overwhelming" feedback head-on. Needs a decision on where the LLM call runs (backend with a configured provider/key vs client-side).
  • MCP-driven import: the connected AI client (e.g. Claude) reads the source doc and calls create_frame (#51) with structured slots. No new LLM infrastructure in Frames; the client's model does the structuring. Good for users already wired to the MCP endpoint.

Guardrails

  • LLM output must validate against the fixed Frame schema (unknown keys rejected).
  • Always show a review/confirm step before anything is created. Never auto-publish unreviewed LLM output.
  • Map only what is confidently present; leave slots empty rather than inventing content.

Dependencies

  • MCP path depends on #51 (create_frame).
  • Web path can reuse the existing PublishFrame RPC.

Acceptance criteria

  • A user can go from a blob of unstructured context to a reviewed, schema-valid Frame without hand-filling every slot.
  • The drafted Frame is presented for review/edit before creation.
  • Output always validates against the schema; low-confidence slots are left empty, not fabricated.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by comparing the proposed web flow with the MCP path, then inspect the existing PublishFrame RPC, the fixed Frame schema, and issue #51's create_frame dependency. Done means unstructured input produces a schema-valid draft that users can review and edit before creation, with uncertain slots left empty.

Written by the indexing model from the issue text.

Assessment

Domain
ai, api, backend, web-dev
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.