meshy-3d-agent
Meshy AI 3D generation and 3D printing skills. Install the whole suite via the meshy-3d-agent plugin, or each skill individually. Version is pinned by .claude-plugin/plugin.json; entries intentionally carry no version field.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Check it deeper on your machine
It has changed since its scan on 12 Nov 2026.
As a Claude Code plugin, install it and then ask Claude Code:
/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/meshy-dev/meshy-3d-agent with scanskillsafety
Or from a terminal, clone it and run it:
git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/meshy-dev/meshy-3d-agent'
It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.
Badge for your README
Markdown
[](/listings/meshy-dev--meshy-3d-agent)HTML
<a href="/listings/meshy-dev--meshy-3d-agent"><img src="/api/listings/meshy-dev--meshy-3d-agent/badge.svg" alt="findsafeskills safety rating" /></a>Install
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
.claude-plugin/marketplace.json · Manifest · 30 lines
- {
- "name": "meshy-3d-agent",
- "owner": {
- "name": "Meshy",
- "url": "https://github.com/meshy-dev"
- },
- "description": "Meshy AI 3D generation and 3D printing skills. Install the whole suite via the meshy-3d-agent plugin, or each skill individually. Version is pinned by .claude-plugin/plugin.json; entries intentionally carry no version field.",
- "plugins": [
- {
- "name": "meshy-3d-generation",
- "source": "./",
- "skills": ["./skills/meshy-3d-generation"],
- "description": "Generate 3D models, textures, and images; rig and animate characters via the Meshy AI API.",
- "author": {
- "name": "Meshy",
- "url": "https://github.com/meshy-dev"
- }
- },
- {
- "name": "meshy-3d-printing",
- "source": "./",
- "skills": ["./skills/meshy-3d-printing"],
- "description": "3D-print Meshy-generated models: slicer detection, white and multicolor printing, Creative Lab products.",
- "author": {
- "name": "Meshy",
- "url": "https://github.com/meshy-dev"
- }
- }
- ]
- }
README.md · README · 212 lines
- # Meshy 3D Agent
- Meshy skills for creating digital assets and preparing 3D prints. All three skills drive
- **Meshy CLI 0.8** (0.8.5 or later, `meshy`) on **Node.js 22.12+**. They are instructions and references only —
- no bundled runtime scripts. Current release: **0.9.6**.
- **Requires a paid Meshy plan (Pro or above),** and generation spends Meshy credits. Sign in with
- your Meshy account; no API key needed. A Free account can sign in, but creating tasks is refused
- (HTTP 402 `plan_upgrade_required`, CLI exit 9). Plans and prices: <https://www.meshy.ai/pricing>.
- ## Let your agent install it
- Paste this into Claude Code, Codex, Cursor, OpenClaw, Hermes Agent, Meta Muse or another agent:
- > Help me install Meshy CLI and the meshy-3d-agent skills:
- > https://raw.githubusercontent.com/meshy-dev/meshy-3d-agent/main/INSTALL.md
- [INSTALL.md](INSTALL.md) walks the agent through the CLI, the skills for its host and one
- browser sign-in.
- ## Start by asking for what you want
- Once installed, say the thing, in your own words:
- > Turn this photo into a textured GLB and save it to `./assets/chest.glb`. Set it up if it
- > isn't yet.
- The agent resolves the CLI, checks whether a session already exists and, if not, starts one
- browser login and shows you a link and a code. You approve in the browser; nothing is pasted
- into the chat, no API key is copied, and the agent then continues the request you actually
- made. The file lands at the path you named — `./meshy_output` is only the fallback when you
- name none — with a rendered preview alongside it and the task IDs for whatever you ask next
- ("make a 1500-face LOD", "now as FBX", "scale it to 150 mm").
- Two more things worth knowing:
- - **The agent installs the CLI.** With Node 22.12+ and no `meshy` (or an older one), it runs
- `npm install -g meshy-cli@0.8`. Where a global install is not allowed, or a newer CLI is
- already installed, it runs the pinned package temporarily instead
- (`npm exec --yes --package=meshy-cli@0.8.5 -- meshy …`).
- - **Local print work needs no account.** Rescaling an OBJ or opening a 3MF in a slicer runs
- entirely on your machine — no login, no balance check, no credits.
- ### Doing it by hand instead
- ```bash
- npm install -g meshy-cli@0.8
- meshy --version
- meshy auth login
- meshy auth status --format json --no-update-check
- ```
- Run `meshy auth login` from a desktop terminal and approve in your browser; in SSH, CI or
- container sessions use `meshy auth login --device` and open the verification URL it prints.
- Keep the login process alive until you have approved. The CLI stores and refreshes the session;
- both skills reuse it under the same OS user and CLI configuration, in any directory. On the China
- site add `--region cn` to the login; the CLI remembers the region for later commands.
- An existing `MESHY_API_KEY` overrides the stored browser session — keep it if that is
- intentional; `auth status` reports the effective source. Explicit key files need
- `--api-key-file`; the CLI never searches `.env`. Never paste a key into a chat. The
- self-contained [setup reference](skills/meshy-3d-generation/references/setup.md) has the
- version, path and error handling in full.
- ## Skills
- | Skill | Workflow |
- |---|---|
- | [meshy-3d-generation](skills/meshy-3d-generation/SKILL.md) | Text/image/2D/motion, textures, remesh, conversion, sizing, UV, rigging and animation |
- | [meshy-3d-printing](skills/meshy-3d-printing/SKILL.md) | White models, multi-color 3MF, analysis/repair, Creative Lab products and slicers |
- Printing installs independently and controls its generation parameters from the start.
- Requested formats and already-approved budgets are preserved. Cost estimates come from
- `meshy make --dry-run` or the published price list, never from your balance; the real charge is
- the task's own `consumed_credits`. No live spending happens during install.
- Install these directories with your host's skills installer, or copy each **whole directory**:
- - Claude Code: `.claude/skills/` (the repository also retains its Claude plugin manifest).
- - Cursor: `.cursor/skills/`, or the whole repository as a plugin (see [Cursor plugin](#cursor-plugin)).
- - Codex: `.agents/skills/`.
- Replace a prior managed skill directory rather than merging files, so old `scripts/` files are
- removed. Back up personal edits before replacement. Do not remove model projects or CLI state.
- Host discovery and browser interaction require a real host smoke test; local contract tests
- alone do not certify a host or operating system.
- ### Cursor plugin
- The repository is also a Cursor plugin. [`.cursor-plugin/plugin.json`](.cursor-plugin/plugin.json)
- loads `meshy-3d-generation` and `meshy-3d-printing`. To load it from a clone instead of the
- marketplace:
- ```bash
- git clone https://github.com/meshy-dev/meshy-3d-agent.git ~/.cursor/plugins/local/meshy-3d-agent
- ```
- Restart Cursor or run **Developer: Reload Window**, then check under **Customize** that both
- skills are listed. Clone into that folder rather than symlinking to it: Cursor skips a symlink
- that points outside `~/.cursor/plugins/local`. On Enterprise plans an admin has to allow local
- plugin imports first.
- You need Node.js 22.12+ and the paid Meshy plan described at the top of this page. The skills
- quote an estimate in credits and stay within the budget you approve before starting a paid task.
- Cursor's agent runs terminal commands in a sandbox that blocks most network access, so the first
- sign-in or task asks to run outside it. To let the Meshy CLI through, add Meshy's domains to
- `.cursor/sandbox.json` in your project (or `~/.cursor/sandbox.json` for every project):
- ```json
- {
- "networkPolicy": {
- "allow": ["api.meshy.ai", "www.meshy.ai", "assets.meshy.ai"]
- }
- }
- ```
- On the China site, use `api.meshy.cn`, `meshy.cn` and `assets.meshy.cn` instead. A team admin can set the same
- allowlist for everyone in Cursor's dashboard. See
- [Cursor's sandbox reference](https://cursor.com/docs/reference/sandbox).
- ### OpenClaw and Hermes Agent
- OpenClaw installs one skill per listing, so it gets both workflows in a single skill that lives in
- [meshy-openclaw](https://github.com/meshy-dev/meshy-openclaw) and carries this repository's
- references. It is published on ClawHub as
- [Meshy 3D Agent](https://clawhub.ai/arlieeee/skills/meshy-3d-agent):
- `openclaw skills install @arlieeee/meshy-3d-agent`.
- Hermes Agent reads the same skills from skills.sh, ClawHub or GitHub:
- `hermes skills install meshy-dev/meshy-3d-agent/skills/meshy-3d-generation`, or
- `npx skills add … -a hermes-agent` as in [INSTALL.md](INSTALL.md).
- Until the 0.6.0 skills release the OpenClaw skill was a separate Python client that called the
- REST API with an API key. It is gone; its task IDs remain usable from the CLI.
- ## Data and local effects
- The CLI sends prompts, media and task requests to the resolved Meshy API origins (normally
- `api.meshy.ai`, or `api.meshy.cn` on the China site); OAuth opens Meshy's browser page. Asset downloads and input URL preflights
- contact their media hosts without the API credential. npm is contacted when the CLI is installed
- or run as a temporary package; workflow commands disable the update notifier. An unexpected
- custom API or login origin must be resolved before credentials are sent.
- CLI profiles normally live in `~/.config/meshy/credentials.v2.json` with mode 0600, their keys
- and tokens encrypted under a key the OS keeps (Keychain, DPAPI or the Secret Service).
- `MESHY_CONFIG_DIR` moves the config root, including operation and device-flow state;
- `MESHY_CREDENTIALS_PATH` overrides only the credential file. The skills never read or copy tokens,
- and never print a token, device code or signed URL.
- `auth use` changes the shared active profile, so concurrent tasks should not switch accounts
- independently. Profile sharing is local, not automatic across machines or containers.
- Models, previews and the small `_meta.json` sidecar that records each delivered task are written
- inside one workspace per job: the directory you named, or `./meshy_output` when you named none.
- Every writing command carries that boundary explicitly, so a path outside it — including through
- a symlink — is refused rather than written. Downloads are selected, not fetched in every format;
- existing files are not overwritten, and `download` replaces one only with `--overwrite`. OBJ preparation writes a new file. Slicer integration launches a detected
- installed application; a launch request is not proof the model was imported.
- ## Development and validation
- Only maintainers need Python 3.11+, PyYAML and jsonschema. None of these tools ships in the CLI
- skill folders.
- ```bash
- python3 scripts/build.py
- python3 scripts/build.py --check
- python3 scripts/validate_skills.py
- MESHY_CLI_BIN=/absolute/path/to/meshy node --test tests/*.test.mjs
- ```
- `build.py` copies the shared setup, delivery and troubleshooting references from generation to
- printing so each skill directory installs on its own; `--check` fails when a copy is stale.
- `validate_skills.py` checks frontmatter, link closure, the absence of bundled runtime files, the
- JSON envelope and resolved workspace placeholder on every documented CLI command, and the plugin
- manifests, for all three skills. It also checks the Cursor manifests against Cursor's plugin and
- marketplace schemas, downloaded from a pinned commit of cursor/plugins and checked by hash (set
- `CURSOR_SCHEMA_CACHE` to keep them in a folder of your choice), and requires the plugin manifest
- to list exactly the generation and printing skills.
- Tests use a loopback API, synthetic credentials and temporary config/workspace directories.
- `tests/runner-contract.test.mjs` additionally runs the documented `npm exec` entry point, so it
- populates an isolated npm cache from the registry on its first run and needs no `MESHY_CLI_BIN`.
- CI installs the pinned CLI into a temporary prefix; tests never contact production or spend
- credits. Without `MESHY_CLI_BIN`, the CLI contract tests skip explicitly — a skip is not a pass,
- so set it.
- ## Upgrade and rollback
- Old task IDs remain usable: `get` or `wait` on the original resource by its ID; never recreate a
- task just to migrate. CLI 0.7.0 removed the project store (`meshy project`, `--project` and `--stage`
- are gone), so a `metadata.json` or `history.json` left by an older version is not read: take the
- task IDs from it by hand. Returning to an old Python skill requires its old runtime and API-key
- setup; it cannot use the CLI OAuth store. Keep the CLI for inspecting in-flight tasks rather than
- submitting again.
- ## Terms, privacy and support
- Using Meshy through these skills is covered by Meshy's
- [Terms of Use](https://www.meshy.ai/terms-of-use) and
- [Privacy Policy](https://www.meshy.ai/privacy-policy).
- [Data and local effects](#data-and-local-effects) lists what the CLI sends and writes. For account
- or billing help, email support@meshy.ai. Report problems with the skills in
- [GitHub issues](https://github.com/meshy-dev/meshy-3d-agent/issues).
- ## License
- [MIT](LICENSE)
- This repository contains only skills and references; it does not include the Meshy CLI or any of
- its dependencies. Setup installs the [Meshy CLI](https://www.npmjs.com/package/meshy-cli) from npm
- as a separate tool, under its own MIT license.
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.