"Guide Wren GenBI applications, MCP server setup, browser-side
Scanned 9/9/2026
Install to Claude Code
npx -y skills add VectorSpaceLab/AREX-Skill --skill genbi-mcp-wasm --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Genbi Mcp Wasm?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/vectorspacelab-genbi-mcp-wasm)More formats (shields.io, HTML) on the badges page.
---
name: genbi-mcp-wasm
description: "Guide Wren GenBI applications, MCP server setup, browser-side
wren-core-wasm APIs, static app verification, and safe Vercel or Cloudflare
deployment."
disable-model-invocation: true
metadata:
disco-role: operating
license: NOASSERTION
---
# GenBI, MCP, and WASM
Use this sub-skill to expose a prepared Wren project to an agent client, build a
browser-side analytics application, or run Wren's browser engine directly.
## GenBI Workflow
1. Confirm a project and its MDL exist. `wren genbi build` can compile a missing
target only when source project files are valid.
2. Choose data mode:
- `snapshot`: bundle small Parquet/DuckDB assets and query in the browser.
- `live`: call a user-owned endpoint at view time; never ship credentials.
3. Request a project-aware build instruction:
```bash
wren genbi build sales-overview --prompt "Revenue by month" --data-mode snapshot
```
4. Write the app only under the generated app directory, copy `mdl.json`, and
include the required data asset for snapshot mode.
5. Register and verify before preview or deployment:
```bash
wren genbi register sales-overview --data-mode snapshot
wren genbi verify sales-overview
wren genbi open sales-overview
```
6. Deploy only when asked, and confirm before `--prod`:
```bash
wren genbi deploy sales-overview --provider vercel
```
## MCP Workflow
Install the extra and build the project first:
```bash
pip install "wrenai[mcp]"
wren context build
wren serve mcp
```
Use stdio when an MCP client launches `wren` itself. Use local Streamable HTTP
when a local client connects to a service:
```bash
wren serve mcp --transport http --host 127.0.0.1 --port 8080
```
Use `--no-connect` for planning/schema/knowledge tools without database query
tools. Use `--allow-write` only when the client should be able to store a query
pair.
## WASM Workflow
Use `@wrenai/wren-core-wasm` for browser-side semantic queries over static or
registered data. Register local data before `loadMDL` in local mode, then use
`query()` or `cubeQuery()`. It is not the server-side connector runtime and has
no semantic-memory module.
## References and Helper
- Read `references/genbi-workflows.md` for data modes, verification, and deploy
decisions.
- Read `references/mcp-server.md` for flags, tools, resources, and security.
- Read `references/wasm-sdk.md` for TypeScript API and browser constraints.
- Read `references/troubleshooting.md` for secret, target, provider, MCP, and
WASM failures.
- Run `scripts/check_genbi_app.py --app <directory> --data-mode snapshot` before
deployment. It is a static check only.
## Guardrails
- A static app must not ship `.env` files, DB passwords, provider tokens, or
connection strings. Treat the built-in verifier as defense in depth, not a
replacement for safe design.
- A successful provider response does not guarantee the deployed URL is publicly
viewable. Check the returned page; Vercel protection can produce 401/403.
- The MCP HTTP transport is local by default and has no bearer-token layer in
this workflow. Do not expose it broadly without an external security design.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!