## Thinking Path > - Paperclip helps people manage AI agents and their work. > - People need consistent UI patterns to review and control that work. > - The current board uses related layouts and controls with different spacing, type, and behavior. > - That drift makes it hard for a designer or an agent to build a new page from a reliable reference. > - Storybook can show the shared patterns and the states they need to support. > - This pull request adds design proposals, a local sandbox, and reusable UI controls so people can compare the patterns in context. > - The benefit is a clearer design system that agents can use to build consistent UI. ## Linked Issues or Issue Description **Subsystem affected** `ui/` — React board UI and Storybook. **Problem or motivation** Board pages repeat visual and interaction decisions. Storybook did not provide a complete reference for task views, conversation states, typography, toolbars, tabs, and light/dark colors. People could not compare a proposed pattern with the running board in one place. **Proposed solution** Add a staged Design Proposals catalog with foundations, components, layouts, and page specimens. Add an opt-in sandbox. Reuse production components where possible. Include synthetic conversation states and shared controls so reviewers can compare and refine them. **Alternatives considered** Separate static mockups would not show keyboard behavior, responsive layout, or real component states. Rebuilding each page only in Storybook would allow the reference to drift from the board. **Roadmap alignment** This improves existing board UI and Skills Manager surfaces. It does not add a new control-plane capability. ## What Changed - Added dark and light color references, typography and conversation specimens, and component stories for buttons, tabs, toolbars, sidebar, popovers, activity, and task lists. - Added a local design sandbox with synthetic task conversation playback and representative states. - Shared collection toolbars and navigation patterns across task, agent, project, routine, skill, connector, and artifact pages. - Added the streamlined Skills tabs and a My Skills collection that uses the shared skill cards. - Restored saved-thread access during synthetic playback, kept connector searchbox semantics, and updated reproduction checks after moving trial stories. - Documented the proposed patterns and added focused UI, catalog, and browser checks. ## Verification - `pnpm check:token-gates`, `node scripts/check-maps.mjs`, and `node scripts/storybook-catalog.mjs --check` passed. - Full workspace `pnpm -r typecheck` and `pnpm build` passed after merging master. After the review fixes, focused UI typecheck, `pnpm build-storybook`, 23 Skills Sources tests, and all 15 Storybook typography browser tests passed. - Focused task-thread and system-notice suites passed: 251 tests. Connector Browse passed 84 tests; the connector, attachment, recovery, and Storybook flyout browser journeys passed locally. - A local full Vitest run hit one intermittent chat-channels integration failure; that test passed in isolation. GitHub CI [39347996059](https://github.com/paperclipai/paperclip/actions/runs/39347996059) passed on the exact latest commit: 55 successful checks and two skipped draft-only Storybook jobs. - Greptile reviewed the exact latest commit at 5/5 with no new comments or open review threads. GitHub reports the branch as cleanly mergeable. ## Risks - This PR changes 206 files and exceeds the repository's 100-file review target. Review by commit and consider a scoped split before merge. - Design Proposals are staged references. Fixture interactions do not prove API behavior, permissions, or provider execution. - UI changes to shared controls may affect pages outside the proposal sandbox. Focused browser checks and the full CI matrix passed. ## Model Used - OpenAI Codex, GPT-6 family. The exact runtime model ID and context window are not exposed in this session. The model used reasoning, code execution, browser inspection, and local tests. Earlier branch commits were also AI-assisted in this Codex conversation. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details available in this session) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above where applicable - [x] I have either linked existing issues or described the issue in-PR following the relevant issue template - [x] I have not referenced internal or instance-local Paperclip issues or links - [x] My branch name describes the change and contains no internal ticket id - [x] I have run focused tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented risks above - [x] All Paperclip CI gates are green - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge --------- Co-authored-by: Scott Tong <scott@scotts-mbp-m5-max.local> Co-authored-by: Paperclip <noreply@paperclip.ing>
Storybook map
Browse the published feature guide,
or run pnpm storybook for the current checkout. The feature and capability IDs
come from the feature recipes. Product behavior and
permissions remain governed by the implementation spec and access companions.
Authored sources
Each story owns its small parameters.paperclip annotation. There is no authored
central catalog, review ledger, date stamp, source list or fingerprint to update.
Storybook's CSF parser supplies filenames, titles, explicit meta IDs, exports,
play-function counts and story IDs without executing story modules.
const meta = {
id: "tasks-create",
title: "Tasks and conversations/Task creation and lifecycle/Create",
parameters: {
paperclip: {
capabilities: ["tasks/create"],
implementation: "production",
boundary: "fixtures",
notes: "Uses the shipped composer; submission is simulated.",
},
},
};
capabilities is an array of stable area/capability IDs. Its first entry is the
primary home; others cross-list the same suite. An empty array requires an
exception explaining the intentional scope. Shared foundations may use
home: "foundations" with an exception to keep their existing location and links.
Staged design work may use home: "design-proposals", with status: "proposed"
and an exception explaining why it is staged separately from production references.
The metadata must be literal; place it after any other parameter spreads.
| Field | Values and purpose |
|---|---|
status |
current, proposed, historical, mixed, unreviewed (default). Authored product alignment intent, not a live review certificate. |
implementation |
production, prototype, mixed, guide, unreviewed (default). What actually renders. |
boundary |
fixtures, fixture-server, static, unreviewed (default). What is simulated or requires an owned fixture. |
notes |
Optional durable scenario limits and rationale. Do not append a history of unrelated changes. |
replacement |
Optional repository-relative path of an existing replacement story. |
A production component can use fixtures. A prototype that imports production
primitives remains a prototype of its outer subject. Use unreviewed when intent
or parity is unclear. Preserve useful historical examples and their explanations.
Historical examples, proposals and guides do not remove missing-example gaps.
No annotation or story count proves permissions, persistence, provider behavior,
full recipe coverage or passing tests.
Navigation and outputs
Stories live in ui/storybook/stories/<primary-feature-area>/. Sidebar titles use
the recipe's section and H1 title, followed by the suite name. / in feature
labels becomes , in the sidebar. Preserve explicit CSF IDs across title/folder
changes. Compare the built index.json when changing navigation.
Design proposals is a separate top-level review section. Its stories live in
ui/storybook/stories/design-proposals/, with titles beginning Design proposals/.
Group these titles by Foundations (tokens), Components (reusable UI),
Layouts (page structures), Pages (complete screens), Patterns (shared
behavior) and Guidelines (usage and contribution rules), in that order.
Create a group when there are stories for it; do not add empty placeholders.
For example, color tokens belong in Design proposals/Foundations/Colors (dark mode),
the task table in Design proposals/Components/Task list, and its complete page
in Design proposals/Pages/Tasks. Preserve explicit story IDs and exports when
reorganizing so links and browser tests remain valid.
Retain capability references so proposals remain discoverable from the feature
guide, but never count them as current production examples. Keep using shared
components and tokens rather than creating a parallel design implementation.
When a proposal is reviewed and adopted in the app, move its story into the usual
feature or Foundations home, update its classification and staging metadata, and
preserve the explicit CSF ID and exports so review links remain valid.
The generator discovers feature recipes and story annotations. It assembles:
ui/storybook/generated/catalog.jsonandfeature-index.json.- A generated Overview story per feature, retaining
feature-map-<area>--overview. ui/storybook/.storybook/preview.tsx, with the literal sidebar order required by Storybook and the authored decorators/fixtures frompreview-base.tsx.- Markdown feature, access, Storybook and verification indexes, plus the current
page inventory, under
maps/generated/.
All these outputs are ignored. Storybook start/build and UI typechecking generate from a fresh checkout. The published Storybook includes the guide and each feature's verification scope. After edits during a dev session, rerun the generator (restart Storybook after navigation changes). Obsolete generated Overview files are removed. Never edit generated outputs or commit them to resolve a merge conflict.
Checks and review
node scripts/check-maps.mjs
node scripts/storybook-catalog.mjs
node scripts/storybook-catalog.mjs --check
node --test scripts/check-maps.test.mjs scripts/storybook-catalog.test.mjs scripts/storybook-feature-index.test.mjs
--write is a compatibility alias for the default generation command. --check
validates authored inputs without requiring existing outputs or writing files.
Checks reject malformed annotations, unknown capabilities, duplicate IDs, broken
map references, missing companions and invalid navigation. CI runs the map tests
through the UI Vitest suite and validates/generates inputs during UI typechecking.
Use Git diffs to select changed stories, rendered subjects, fixtures, and recipes for review. A changed source is a review candidate, not an instruction to refresh metadata on every dependent example. Inspect actual imports and contracts; shared fixture changes can require broad review without broad acknowledgement edits. Keep revision-specific findings and executed evidence in the PR, issue or CI artifact. Move only durable limitations into the affected source annotation or recipe. Past review history remains available through Git.
maintain-storybook owns stories and their annotations; review-storybook exercises
them. Product/access maintenance owns the recipes, including optional capability
verification scope. Use the existing Storybook browser workflow.
Report source inspection, rendering, play assertions, visual comparison and
production acceptance separately. Do not publish new baselines to pass a map cleanup.