Hacktoberfest 2026: the issues maintainers tagged for October, open and beginner-friendly. Browse Hacktoberfest issues

P5.3: Web-grounded agent entry point — llms.txt + published agent knowledge on ui.mieweb.org

Open
  • Is something wrong?
#626 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
5/5
Estimated time
Over a week
Newbie friendliness
25/100
Issue type
Feature
Clarity
Mostly clear
Activity status
Active
Tech stack
markdown, storybook
Domain
developer-experience, documentation

Research direction

Start with how catalog-manifest.json is already served through Storybook staticDirs, then read src/Introduction.mdx for the landing page. The P1 skill generators (#589, #600, #579) must exist before the markdown output target can be added, so check their status first. Done means llms.txt resolves at the site root, the generated /agent/*.md files rebuild with the site, the landing page links to them, and an eval question is added under agent/evals/.

Written by the indexing model from the issue text.

Description

Part of #545 (Phase 5). Depends on the P1 skill generators (#589, #600, #579); best done alongside P5.2 (#607).

Use case

A business analyst (or anyone) tells a generic, web-capable AI agent: "I'm making [an app/slide/screen] and was told to use https://ui.mieweb.org — build this to the MIE UI." No repo, no installed dev env, no init-agent. Today that agent freestyle-scrapes JS-heavy Storybook HTML and gets poor results. The site should instead carry the design-system knowledge in agent-findable, agent-readable form.

Scope

  1. llms.txt at the site root (ui.mieweb.org/llms.txt, per the llmstxt.org convention): identifies the design system and links to the machine-readable knowledge below. Served via Storybook staticDirs (the site is the built storybook-static/; catalog-manifest.json is precedent).
  2. Publish generated agent knowledge as static markdown on the site (e.g. /agent/catalog.md, /agent/tokens.md, selection/usage guidance, key policies: NITRO for tables, no raw-HTML patterns, Lucide icons). Reuses the P1 skill generators with a second output target — single source of truth, no drifting copy (epic principle 5; extends principle 8's tool-agnostic distribution).
  3. "For AI agents" section on the landing page (src/Introduction.mdx): short visible block pointing agents at /llms.txt, and telling repo-based users/agents to run npx @mieweb/ui init-agent for the full team.

Known limitation (by design)

A web-grounded agent gets the orchestrator's knowledge but not the team — no subagents, handoffs, or auditor tool restrictions. That's acceptable: it upgrades ad-hoc scraping into grounded guidance, and funnels repo-based work toward the real install.

Done when ui.mieweb.org/llms.txt resolves, links to generated markdown that regenerates with the site build, the landing page carries the agent pointer, and a clean agent session given only the site URL produces on-system component selection (eval question added to agent/evals/).

Dominant language
TypeScript
Stars
2
Forks
4
Avg merge
1d 2h
Merged PRs (30d)
72

Getting set up

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.

More from mieweb/ui

All issues in mieweb/ui

Similar issues

More TypeScript issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.