# zcag/tela [Health: Active]

**Category:** 🧠 Knowledge & Memory  
**Repository:** https://github.com/zcag/tela  
**GitHub Stars:** 89  
**npm Downloads (last month):** 168  
**Views:** 4  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/zcag-tela

## Description
Self-hostable, markdown-native team wiki with a built-in MCP server: agents search, read, and write your wiki pages (ranked Postgres full-text + semantic search, backlink traversal). Plus Atlas, which auto-generates a cited, coverage-checked wiki from your git repos and Jira. Go + Postgres. npx tela-mcp

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "tela": {
    "command": "npx",
    "args": ["-y","tela-mcp"],
    "env": {
      "TELA_PUBLIC_BASE_URL": "",
      "TELA_SHARE_SECRET": "",
      "TELA_API_KEY_SECRET": "",
      "TELA_PG_PASSWORD": ""
    }
  }
}
```

**Requires environment variables:** `TELA_PUBLIC_BASE_URL`, `TELA_SHARE_SECRET`, `TELA_API_KEY_SECRET`, `TELA_PG_PASSWORD` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What zcag/tela MCP server does

The zcag/tela MCP server gives agents access to a tela wiki as both a knowledge source and an authoring surface. It works with tela's markdown-native page store, where page bodies remain canonical Markdown rather than being converted into a proprietary block format. Agents can find relevant pages, retrieve their contents, create documentation, and update existing pages.

The wiki is backed by a Go application and PostgreSQL. Search combines ranked PostgreSQL full-text retrieval with optional semantic search through pgvector. Page relationships can also be explored through backlinks, which helps an agent move from a matching page to related documentation.

## How it works

The MCP interface is built into tela's backend at `/api/mcp`, rather than being a separate add-on to the wiki. An agent connects to the tela instance and operates against the same content that users edit in the browser. Write access can be scoped per tool, allowing deployments to distinguish between agents that may only read documentation and those permitted to author or revise pages.

Full-text search works without an embedding service. Semantic retrieval requires an embedder and pgvector; the bundled deployment can run Ollama as an optional embedding service. Atlas is tela's documentation engine for generating and maintaining wiki pages from sources such as Git repositories and Jira, complementing the direct MCP workflow.

## Setup and configuration

A self-hosted tela deployment uses Docker Compose and Make. Clone the repository, run `make setup` to create the environment file and secrets, set the public URL, then start the stack with `make up`. The service is exposed through Caddy on port 8780 by default, and the first launch provides a setup wizard unless bootstrap admin credentials are configured.

The deployment requires a public base URL, share-link signing secret, API-key signing secret, and PostgreSQL password. Keep the signing secrets stable: changing them invalidates existing share cookies or personal access tokens. The `tela-mcp` npm package can also be launched with `npx tela-mcp`.

## Tools and capabilities

- Search wiki content using ranked PostgreSQL full-text search.
- Use optional semantic retrieval backed by pgvector and an embedding service.
- Read Markdown page content from the connected tela wiki.
- Create and update wiki pages through agent-authorized write tools.
- Traverse backlinks to discover related pages.
- Work against the same pages maintained by browser users and Atlas.

## Limitations and notes

Semantic search is optional; full-text search is the default retrieval path. Enabling semantic retrieval requires pgvector and an embedder, such as the optional Ollama service in the Compose setup. Write operations depend on the permissions granted to the connecting agent, so a read-only configuration will not support page authoring.

Self-hosting requires Docker with Compose and Make for the bundled deployment. Removing Docker volumes with `make clean FORCE=1` deletes stored data, and rotating the API-key secret invalidates existing personal access tokens.

_Full upstream README: https://allmcps.com/mcp/zcag-tela/readme_