ProfRandom92/comptext-mcp-server
If you are the rightful owner of comptext-mcp-server and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to dayong@mcphub.com.
CompText MCP Server is a token-efficient domain-specific language server designed for seamless LLM interactions across multiple AI platforms.
🚀 CompText MCP Server
Token-efficient DSL server for MCP, REST, and mobile agents
CompText compiles verbose instructions into a compact DSL to cut LLM token usage by up to 90–95%. This repository ships a production-ready MCP server, REST gateway, and a mobile automation agent.
Table of Contents
- Overview
- Features
- Architecture
- Prerequisites
- Quick Start
- Configuration
- Usage
- Testing & Linting
- Documentation
- Contributing
- Security
- License
Overview
CompText MCP Server exposes the CompText DSL through:
- MCP for native tool access (Claude Desktop, Cursor, etc.)
- REST for HTTP clients (Perplexity, ChatGPT API-style callers)
- Mobile agent for Android automation with minimal tokens
All services share the same validation, caching, metrics, and security hardening described in .
Features
- Token-efficient DSL with caching and input validation
- Dual interfaces: MCP server and REST wrapper
- Mobile agent with Ollama/Cloud modes and Prometheus metrics
- Rate limiting, sanitization, and structured logging
- Docker support plus Railway/render configs
- GitHub repository automation: Audit, auto-merge PRs, and manage default branches
Architecture
Client (MCP / REST / Mobile) -> CompText Server -> Data Source (Local JSON or Notion)
|-> Validation & rate limiting
|-> Metrics & logging
|-> Caching layer
Data Sources
The server supports two data source modes:
- Local JSON (default): Uses
data/codex.jsonfor fast, offline access - Notion API: Uses Notion database for cloud-based content management
Switch between modes using the COMPTEXT_DATA_SOURCE environment variable.
Prerequisites
- Python 3.10+
- (Optional) Notion API token if using Notion as data source
- Recommended: virtualenv
Quick Start
git clone https://github.com/ProfRandom92/comptext-mcp-server.git
cd comptext-mcp-server
python -m venv .venv
source .venv/bin/activate # Windows: .venv\\Scripts\\activate
pip install -e .[rest,mobile]
cp .env.example .env
# Option 1: Use local JSON (default, no additional config needed)
python -m comptext_mcp.server
# Option 2: Use Notion API (edit .env first)
# Set COMPTEXT_DATA_SOURCE=notion
# Fill in NOTION_API_TOKEN and COMPTEXT_DATABASE_ID
python -m comptext_mcp.server
# (Optional) Start REST API
python rest_api_wrapper.py
Configuration
All configuration uses environment variables (see ):
Data Source Configuration
COMPTEXT_DATA_SOURCE– Data source: "local" (default) or "notion"COMPTEXT_CODEX_PATH– Path to local JSON file (default: "data/codex.json")
Notion API Configuration (only if using Notion)
NOTION_API_TOKEN– Notion API tokenCOMPTEXT_DATABASE_ID– Notion database ID
Other Configuration
GITHUB_TOKEN– GitHub API token (for automation features)HOST/PORT– REST server host/portLOG_LEVEL– Logging level
Usage
MCP Server
Add to your MCP client (example for Claude Desktop):
Using Local JSON (default):
{
"mcpServers": {
"comptext-codex": {
"command": "python3",
"args": ["-m", "comptext_mcp.server"],
"cwd": "/path/to/comptext-mcp-server",
"env": {
"PYTHONPATH": "/path/to/comptext-mcp-server/src",
"COMPTEXT_DATA_SOURCE": "local",
"COMPTEXT_CODEX_PATH": "data/codex.json"
}
}
}
}
Using Notion API:
{
"mcpServers": {
"comptext-codex": {
"command": "python3",
"args": ["-m", "comptext_mcp.server"],
"cwd": "/path/to/comptext-mcp-server",
"env": {
"PYTHONPATH": "/path/to/comptext-mcp-server/src",
"COMPTEXT_DATA_SOURCE": "notion",
"NOTION_API_TOKEN": "your_token",
"COMPTEXT_DATABASE_ID": "0e038c9b52c5466694dbac288280dd93"
}
}
}
}
REST API
python rest_api_wrapper.py
curl http://localhost:8000/health
See for endpoints, rate limits, and examples.
Mobile Agent CLI
comptext-mobile run "Open Chrome and search for weather" --steps 10
comptext-mobile status
comptext-mobile screenshot --output screen.png
Configure via environment or a config file; details in .
Testing & Linting
pip install -e .[dev]
pytest
black . && flake8 && mypy
Documentation
Contributing
We welcome issues and PRs. Please see and follow the .
Security
Report vulnerabilities via . The project uses input validation, rate limiting, and CodeQL scanning (see CI).
License
MIT © ProfRandom92