dsh-web-ui
zhu1090093659
Plugin and skin collection for DeepSeek Harness (DSH) Web UI - task board, git graph, right-side panel, remote mobile UI, pet, live token stats, and skin center.
Bandersnatch0x/design-playbook
Design I/O plugin for Claude Code & coding agents — declarations + contracts that make UI generation constrained, reviewable, and recirculatable. Not a style pack; composes with ui-ux-pro-max + frontend-design.
PROJECT TOPICS
INSTALL REFERENCE
dsh plugin --profile web add github:Bandersnatch0x/design-playbook
该命令指向仓库当前默认分支;尚无绑定当前 commit 的完整验证结果。
PROJECT README
Not another style/palette pack. Compose with ui-ux-pro-max + frontend-design; this plugin owns the pipeline and acceptance.
A Claude Code / Codex plugin. One predictable pass per run — Design I/O: design-baseline? → reference-intake? → ux-spec? → plan? → (native-craft?) → ui-picker → (preview*) → fill → craft-guard → (observe*) → ui-evaluator, where acceptance points back to the declaration that owns each failure, and blocking findings recirculate until closed. ? = conditional entry (design-baseline? for UI work in an existing product; reference-intake? when screenshot/URL/design/product analogy is present); preview*/observe* run only when their optional MCP adapter is present (preview_prototype before Fill, execute_capture_plan after craft) — otherwise skipped. preview* is a human-in-the-loop confirm gate (G5); observe* captures criterion-addressable runtime evidence into a manifest the evaluator binds to a criterion (G6).
spec · domain · craft · design · components · templateskill (timing) · evaluator (acceptance + recirculate)🎬 Try it:
/design-playbook:design-io <your ask>— one pass landsspec.md, a decision report, and a point-back ledger under.scratch/<run>/(artifact shape:showcase/01-spec.md). See it run on a real project: theshowcase/folder is a full Design I/O pass against SwarSight — spec, decision report, and point-back critique with a closed recirculate trail.
Claude Code
/plugin marketplace add https://github.com/Bandersnatch0x/design-playbook.git
/plugin install design-playbook@design-playbook
Codex
codex plugin marketplace add Bandersnatch0x/design-playbook
codex plugin add design-playbook@design-playbook
The marketplace catalog lives at the repo root (not the package):
claude --plugin-dir <abs>/packages/design-playbook # dev load, no install
/plugin marketplace add <abs-to-repo-root> # local marketplace
/plugin install design-playbook@design-playbook
codex plugin marketplace add <abs-to-repo-root>
codex plugin add design-playbook@design-playbook
Codex bridge notes: packages/design-playbook/codex/AGENTS.md.
Invoke namespaced: /design-playbook:design-io <ask>. Bare /design-io is a --plugin-dir dev alias only.
Eight model-invoked skills (/design-playbook:<name>):
| Skill | Role |
|---|---|
design-playbook |
🎯 Orchestrator (full pipeline, run-profile tiering P1/P2/P3) |
design-baseline |
🧭 Discover, validate, or draft project DESIGN.md before existing-product UI work |
reference-intake |
📎 Reference contract (screenshot/URL/analogy → Keep/Change/Do not copy) |
ux-spec |
📋 Six-layer spec declaration via the S0-S6 shaping session (question/assumption/confirmation batches + session artifacts) |
ui-picker |
🧱 Shell + component semantics + R/C/E design-decision entries |
craft-guard |
🛡️ Craft / anti-AI-slop against the first-party rule registry (seven-column audit rows) |
native-craft |
🖥️ Native-feel desktop declaration |
ui-evaluator |
✅ Dual-track acceptance (six-block point-back) + recirculate |
Commands: design-io (full pipeline) · ux-spec (spec only) · ui-review (accept only) · run-review (cross-run) · run-status (phase + resume narration) · doctor (install health)
Every run declares a tier in the plan.md run-profile block — process weight stays proportional to change consequence. The agent grades against the declaration-touch checklist, the user confirms once; upgrades are automatic the moment a correction signal appears, downgrades need the user.
| Tier | Scope | Gate face |
|---|---|---|
| P1 point-fix | Single-owning-layer point-back repair, no decided-field touch | Registry subset evaluation; R4/R5 (+R2 line) routes |
| P2 standard | In-baseline feature change (new criteria, R/C decisions) | Full predicate evaluation; shaping session + G9/G10 |
| P3 full | Decided-field revision (supersedes), structural re-composition, E-tier decisions | G1-G12 full spectrum + sampling matrix fully executed |
Full matrix and re-entry semantics: docs/specs/ui-ux-vnext/loop-prototype.md.
| Package | Use for |
|---|---|
| design-playbook | Baseline? → Reference? → Spec? → plan? → shell → optional preview → fill → craft → optional observe → point-back |
| ui-ux-pro-max | Style / palette / type search |
frontend-design |
Anti-template visual direction |
| native-feel-skill | Full native-feel depth (WebView, IPC, memory) |
Preview and Evidence MCP runtimes ship inside the main plugin
(packages/design-playbook/mcp/ + .mcp.json with ${CLAUDE_PLUGIN_ROOT}).
Marketplace install therefore registers both tools without a second package.
The orchestrator still probes and skips steps when a host has no MCP tools.
| Adapter | MCP tool | Enables | Notes |
|---|---|---|---|
design-playbook-preview |
preview_prototype |
preview* human confirm gate (G5) |
Bundled; needs system Edge/Chrome for the popup (falls back to default browser); sibling dir is a compatibility launcher |
design-playbook-evidence |
execute_capture_plan |
observe* runtime evidence (G6) — needs Playwright + Chromium |
Bundled; capture still optional at runtime |
.claude-plugin/marketplace.json ← repo-root catalog (source: ./packages/design-playbook)
packages/design-playbook/ ← public plugin (skills, commands, mcp/, examples, showcase)
packages/design-playbook/mcp/ ← bundled Preview + Evidence MCP runtimes
packages/design-playbook-preview/ ← compatibility launcher + docs (G5)
packages/design-playbook-evidence/← compatibility launcher + docs (G6)
docs/agents/ docs/adr/ ← engineering shell (tracker, workflow, decisions)
CONTEXT.md .scratch/ ← glossary, specs, tickets, dogfood logs
Runs land their artifacts under .scratch/<run>/ in your project — see the package README and SKILL.md steps 3, 5, 8.
Maintainer helpers: scripts/doctor.py (install health), packages/design-playbook/scripts/run_status.py (derive status/resume from run artifacts), scripts/release.py (tag gate), scripts/validate.py (static plugin surface).
Root = GitHub front door + engineering shell · Package = only runtime surface · product-* maintainer commands stay at root, never in the package.
MIT (authored content). See LICENSE + NOTICE. No rights claimed over any third-party playbook corpus.
CLASSIFICATION EVIDENCE
系统优先读取 GitHub Topics,再与站内分类词典和词根规则比对。当前命中: agent-skills、ai-coding、claude-code、claude-code-plugin、frontend-design、ui-design、ui-ux。