Lobe Design Cli
How agents use the Lobe Design CLI (lobe-design / lhd) against design.lobe.li: auth (browser device login or optional DESIGNCLITOKEN), search/list/get for 设计产出 (mock) and 设计灵感 (case), optional CRUD with local media conversion, archive admission gates before create/upload, and lobe-design skill install to wire lobe-design + this skill into a project. Trigger on lhd, lobe-design, lobe-design-cli, design.lobe.li CLI, auth login, DESIGNCLITOKEN, archive search, 设计产出, 设计灵感, skill install, 门禁, curated, STAR.
mkdir -p ~/.claude/skills/lobe-design-cli && curl -fsSL 'https://raw.githubusercontent.com/lobehub/lobehub/canary/.agents/skills/lobe-design-cli/SKILL.md' -o ~/.claude/skills/lobe-design-cli/SKILL.md
Get the complete skill folder, including its scripts and reference files.
When to use this skill
Use Lobe Design Cli when an AI agent needs a reusable SKILL.md workflow for this job: How agents use the Lobe Design CLI (lobe-design / lhd) against design.lobe.li: auth (browser device login or optional DESIGNCLITOKEN), search/list/get for 设计产出 (mock) and 设计灵感 (case), optional CRUD with local media conversion, archive admission gates before create/upload, and lobe-design skill install to wire lobe-design + this skill into a project. Trigger on lhd, lobe-design, lobe-design-cli, design.lobe.li CLI, auth login, DESIGNCLITOKEN, archive search, 设计产出, 设计灵感, skill install, 门禁, curated, STAR.
When not to use it
Skip Lobe Design Cli when the task is outside the productivity category, or when a more specific skill in this directory already covers the same workflow with clearer triggers.
How to install
Where will you use it?
mkdir -p ~/.claude/skills/lobe-design-cli && curl -fsSL 'https://raw.githubusercontent.com/lobehub/lobehub/canary/.agents/skills/lobe-design-cli/SKILL.md' -o ~/.claude/skills/lobe-design-cli/SKILL.md
- Claude Code personal install: place the skill and its bundled files at ~/.claude/skills/lobe-design-cli/SKILL.md. For a shared project, commit the folder at .claude/skills/lobe-design-cli/ instead.
- Current Claude Code detects SKILL.md changes in watched directories during the session. If you just created a top-level skills directory, run /reload-skills. Check that the skill is listed before testing its trigger.
- For Claude chat or Cowork in the Desktop app, manage uploaded skills in Customize > Skills; copying files into a local Claude Code folder is not a Desktop install. Other agents may use different install locations.
Full install guide for Claude, Cursor, and Codex →
Try a starter prompt
After installing the skill, paste this starter prompt into your agent to verify it triggers on a small task.
Use the Lobe Design Cli skill for this task: Use Lobe Design Cli when this documented workflow matches the task. Confirm any required inputs or files before making changes.
What supports this recommendation?
71/100 provisional content score. This measures documentation, practical detail and source signals. It is not a measured task success rate or a security certification.
- Documentation
- 77/100
- Practical detail
- 68/100
- Source evidence
- 63/100
- Source signals
- 72/100
Model compatibility evidence
- claudeinferred
The skill text mentions Claude or a closely associated term. This is keyword-derived compatibility; no hands-on model test is recorded.
- chatgptuntested
No model-specific signal or recorded compatibility test was found.
- geminiuntested
No model-specific signal or recorded compatibility test was found.
- copilotuntested
No model-specific signal or recorded compatibility test was found.
- llamauntested
No model-specific signal or recorded compatibility test was found.
- perplexityuntested
No model-specific signal or recorded compatibility test was found.
- mistraluntested
No model-specific signal or recorded compatibility test was found.
- grokinferred
The skill text mentions Grok or a closely associated term. This is keyword-derived compatibility; no hands-on model test is recorded.
Inferred compatibility is based on content signals. Check the app, its available tools and setup requirements before running a skill.
What this skill does
> CLI binary: **`lobe-design`** (short alias **`lhd`**). Do **not** use `ld` — it conflicts with the system linker (`/usr/bin/ld`).
# lobe-design-cli
Operate the **Lobe Design archive** from the terminal. Full design-system pipeline lives in the sibling skill **`lobe-design`**; this skill covers the `lobe-design` tool **and the admission gates agents must pass before mutating the archive**.
Canonical deep dive (also linked from lobe-design): `skills/lobe-design/references/cli.md`. **Full gates:** [references/gates.md](references/gates.md) — read before every create/upload.
## Install skills into a project
```bash lobe-design skill install --dir < project > /.agents/skills # or: --claude / --codex / --cursor / --agents / --all # snapshot instead of symlink: --copy ```
This installs **both** `lobe-design` and `lobe-design-cli`. Equivalent bash entry: repo-root `./install.sh` (now also links this skill when present).
## Auth
Members sign in with `lobe-design auth login` (browser, <https://design.lobe.li>). Previously saved tokens remain valid. Automation may use `--token` or `LD_TOKEN`. Personal device sessions last 30 days and renew on active use after server-side credential refresh and membership checks. First sign in again at <https://design.lobe.li/login> and consent to `offline_access`, then run `lobe-design auth login`. Reauthenticate if refresh credentials are rejected. Static token logins are unchanged.
```bash lobe-design auth login lobe-design auth status lobe-design auth logout
# automation / CI (optional) lobe-design auth login --token "$DESIGN_CLI_TOKEN" ```
## Read / write cheatsheet
```bash lobe-design search "sidebar" --kind all lobe-design list --kind mock --json lobe-design get --json < id > --related
lobe-design create ./entry.json --kind mock --media-dir ./design-archive lobe-design update ./entry.json --kind case --media-dir ./design-archive lobe-design upload ./entry.json --kind mock --media-dir ./design-archive lobe-design delete < id > --yes ```
Kinds: `mock` = 设计产出,`case` = 设计灵感. Default base URL `https://design.lobe.li`. Media: images → WebP, video → WebM via CLI; CDN `https://design.lobeobjects.space`.
## 门禁 / Gates (mandatory before create /upload)
Agents **must** apply these gates before `create` / `upload` / curated `update`. Do not invent HTTP against `/api/trpc` to bypass them. Detail + checklist: [references/gates.md](references/gates.md). Schema: `packages/cli/src/utils/schema.ts`.
### Two kinds (browse UI)
| kind | 中文 | Meaning | | ------ | ------------ | --------------------------------------------- | | `mock` | **设计产出** | Our design work with STAR + tracing + rating | | `case` | **设计灵感** | Curated external / reusable design references |
Browse site shows **only** these two. `raw` / `rejected` stay off the UI.
### 设计灵感 (`case`) — what may enter curated
**Admit** (`tier: curated`, prefer `case_admit: design_ref`):
- Learnable product / UI references; landing / pricing finished comps - Design-method covers / redline annotations; photo→finished design contrasts - Magazine-style design cards (subject = the design)
Quality bar ≈ `#design` / `#case-room`, but **not channel-gated** — other channels OK if same quality.
**Reject** (`tier: rejected` — do **not** curated-upload):
- Requirement / complaint / naming-rant screenshots - Engineering bug dumps; random `#dev` likes ≠ curated - LobeHub own promo (brand bars, promo posters, launch cards) - Agent image / motion workflow demos (chat + preview) - Real-person street / event / community photos; photo-reference raws (subject = person) - Isolated IP / mascot / blind-box stills; pure character marketing - Booth / hardware snaps; unrelated social screenshots
**Fields:** bilingual `title` + `title_en` + tags (UI shows zh); title = object+pattern; tags fine-grained **\~5–12** (object / component / interaction / platform / state / technique), never a single coarse tag like `other`; write **`core`** (true subject, e.g. 本地脱敏) + **`summary`** + emphasize **`why_good`**; `sources` include original URL (X / Discord).
### 设计产出 (`mock`) — gates
- Split by **case dimension** (each exploration / direction = own entry), **not** by Discord thread. Example: Agent 头像「脸型矩阵」vs「状态环」= two mocks. - Title: object + change/mode; bilingual title/tags; fine tags 5–12. - **Final assets:** all directions that are stage-OK (not only last few frames); weak process drafts → **tracing**, not final album. Cover 「n 图」= final album count. - `status`: `shipped` | `iterating` | `rejected` | `superseded`.
### STAR writing (`mock.star`)
Concrete Chinese prose — not empty slogans:
- **situation**: context / constraints / who asked - **task**: design goal / decision to make - **action**: what was tried / chosen (tradeoffs) - **result**: outcome / what shipped or stuck
### 好在哪 / 差在哪
- `problems[]` as `{ problem, solution? }` or strings for gaps/risks (差在哪) - `rating_pros` / `rating_cons` + `rating_reason` for 好在哪 / 差在哪 tied to score - STAR Result may state what worked; problems must not duplicate empty fluff
### Scoring (`rating` 0–5)
- Required on mocks when context exists — never leave blank if Discord human feedback + STAR/problems can support an inference - UI shows **stars only** (no 「5/5」 text) - Always fill brief `rating_reason` - Rubric: **5** strong praise / clear win / few residual problems · **4** solid, minor nits · **3** mixed / iterating with real gaps in problems · **1–2** weak / rejected path / major issues · **0** unusable / wrongly filed
### Effective tracing
- Full related-people chat, not Helper-only; include Discord **and** Grok Bot design dialogue when relevant - Avatars for user + agents; resolve `@` mentions to usernames (no bare ids) - Images sized like Discord chat (not full-bleed); attach process drafts here - No useless phase labels - Prefer `discord.messages` / thread ids / urls (schema already supports)
### Pre-flight (block create/upload if any fail)
1. Kind correct (`mock` vs `case`)? 2. Case: admit → `tier: curated` + fields; else `rejected` + `reject_reason` (no curated upload)? 3. Mock: one case-dimension per entry; final vs tracing split correct? 4. Bilingual title + 5–12 fine tags? 5. Case: `core` / `summary` / `why_good` / source URL? Mock: STAR four fields concrete + problems + rating + `rating_reason`? 6. Tracing useful when claiming a design story? 7. Bins are `lobe-design` / `lhd` only?
## Rules of thumb
1. Use `lobe-design` instead of hand-rolled HTTP to `/api/trpc`. 2. Parse `--json` output; do not scrape TTY tables. 3. Mutate the archive only when the user asked; always pass `--yes` for deletes. 4. **Pass 门禁 / Gates above** (and [references/gates.md](references/gates.md)) before every create/upload. 5. After installing skills, tell the user to run lobe-design `05-init` / bootstrap if the host project has no `.agents/ux/product.md` yet.
Intended uses
- Use Lobe Design Cli when this documented workflow matches the task.
Implementation notes
- Review the source instructions and adapt inputs before running the workflow.
Try it, then tell us how it went
A copied prompt does not tell us whether the task worked. Report your result after checking the output. We collect your tool and outcome, never your files or prompt.