SkillFoxx

Launchstack MCP

Launchstack MCP updates a saved conversation using its previous identifier so a continuation does not create another copy.

  • 897 ★ on the repository
  • Last commit
  • Apache-2.0
MCP server

Extended access

Why this level

  • The connected agent receives documents within workspace permissions
  • Saving a conversation adds or updates a team document
  • Hosted models may receive document and conversation text
All reasons and checks

deodat-lawson/launchstackAuthor:Deodat-Lawson

Install

Manual install

git clone https://github.com/Deodat-Lawson/LaunchStack.git
cd LaunchStack
pnpm install
cp .env.example .env

Prepares the app. Configure authentication secret, origin, model providers and embeddings before starting Docker. Requires Node.js 20 or later, pnpm 10.15.1 and PostgreSQL with pgvector.

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • The connected agent receives documents within workspace permissions
  • Saving a conversation adds or updates a team document
  • Hosted models may receive document and conversation text

This is third-party code. Review the repository files before installing.

What it does

Launchstack MCP connects an agent to documents in its workspace through its built-in server. The agent searches material, fetches a selected source and lists available documents. Its account and workspace permissions determine what is accessible. The separate `save_conversation` tool stores a conversation as a document for later retrieval. The user agrees on the included messages and the workspace where the document will be stored. MCP belongs to the document-management application. Deploy the app and process uploaded material before the first search. Application features use configured model providers; self-hosting does not rule out text reaching an external provider.

Who it is for. Document-heavy teams searching their work archive through an agent and retaining useful conversations alongside it.

Good fit when

  • An agent searches processed team documents for a source.
  • A selected document is retrieved under the user’s permissions.
  • An owner requests that a useful conversation be saved to the team archive.
  • A connection is limited to a particular workspace.

Not a fit when

  • You need a small server without deploying an application and storage
  • You require search without configuring an embedding provider
  • Conversations cannot be stored in a shared team archive

Example request

Find Launchstack documents about this project’s requirements and show sources. Do not save this conversation yet.

MCP

Transport
http
Authentication
OAuth

Environment variables

BETTER_AUTH_URL required
Your installation origin; MCP needs HTTPS, or loopback HTTP for local development.
MCP_SERVER_ENABLED
Set false to disable the application MCP server.

Limitations

Search needs EMBEDDING_API_BASE_URL and EMBEDDING_API_KEY; uploading documents alone does not create a searchable index. Chat uses the configured model, defaulting to Gemini; Azure credentials can route PDFs and images to Azure OCR. Remote MCP requires HTTPS and a correct BETTER_AUTH_URL. Docker memory use depends on configuration and load: estimates are about 400 MB for lite and 1.2 GB with Docling OCR, so test resources with your own documents.

How to disable. Remove the agent connection and revoke its grant or key in your installation. `MCP_SERVER_ENABLED=false` disables application MCP. `make down` stops the deployment while preserving data volumes. Handle deletion of saved conversations separately after reviewing permissions and backups.

FAQ

Where is my Launchstack MCP endpoint?

At `/api/mcp` under `BETTER_AUTH_URL`. External clients use your HTTPS domain; local development permits `http://localhost:3000/api/mcp`.

Is there a standalone npm MCP installation command?

There is no standalone ready-to-run npm server here. Deploy the full Launchstack application, then configure workspace MCP access.

Why is an uploaded document missing from search?

Wait for the background worker to process the file and check embedding configuration. The web app only accepts the upload. Once processing is complete, compare the user’s access with the document permissions.

What controls automatic conversation saving?

Automatic saving depends on client instructions and the access grant. The tool requires documents.upload and launchstack:write. Agree with the owner in advance on which conversations may be saved.

Which messages are saved?

The tool accepts user and assistant turns in order, without summaries or tool output. Agree on sharing before saving private conversations to a team workspace.

What license and model costs apply?

Code uses Apache-2.0; third-party components retain their terms. Models and hosting are billed separately.

Discussion

No questions or reviews yet.

MCP servers
Official

The MCP server lets agents explore prepared documents, sections and related assets while retaining source references.

MCP serverExtended access4K
Official

The MCP server lets agents read and search local Molio vaults, browse file trees and ask questions about their contents.

MCP serverExtended access448

A local document base for your agent: hybrid semantic and keyword search plus a web UI for browsing and uploads

MCP serverExtended access346
Official

The MCP server connects an agent to a learning canvas where a question about a passage opens a linked document.

MCP serverExtended access334
Official

An MCP server that lets an agent read long PDFs through a document tree: retrieval by reasoning instead of a vector database

MCP serverExtended access412

README in short

The app processes documents through separate background jobs. Docker-stack composition determines available recognition options, while model credentials and user permissions are configured independently.

README badge

Report a problem