ToolFront exposes actions and data workflows as MCP endpoints that enable AI agents to execute RAG (Retrieval-Augmented Generation) tasks against both local and remote data sources. It is built on the Statespace platform, a Markdown framework for creating agent-native APIs. Key capabilities of ToolFront: - Execute RAG tasks against local databases, files, and knowledge bases - Connect to remote data sources for agent-driven retrieval - Define tools and workflows using Markdown with YAML frontmatter - Support for text-to-SQL, log analysis, and knowledge base queries - Built-in tool declarations for system commands, Python scripts, and SQLite databases - Deploy applications locally or to the Statespace cloud platform - Create multi-page knowledge systems with structured data and instructions

*A simpler way to build agent-native APIs.*

[](https://github.com/statespace-tech/statespace/actions/workflows/test.yml) [](https://github.com/statespace-tech/statespace/blob/main/LICENSE) [](https://crates.io/crates/statespace) [](https://discord.gg/rRyM7zkZTf) [](https://x.com/statespace_tech)

---

Website: https://statespace.com

Documentation: https://docs.statespace.com

---

Statespace is a Markdown framework for building REST APIs that agents can directly interact with. Build RAG, text-to-SQL, knowledge bases, and more — in pure Markdown. Once you’ve created an app, you can deploy, manage, and share it from our cloud platform.

Installation

Install the CLI:

curl -fsSL https://statespace.com/install.sh | bash

Example

1. Create it

Create a file README.md with:

---
tools:
    - [date]
---

This is an example application.
Instructions
Check the current timestamp with date

2. Run it

statespace serve .

3. Ask it

Pass the URL to your agents:

claude "What can I do with the API at http://127.0.0.1:8000?"

4. Update it

Add data files, scripts, and more Markdown pages to your app:

demo/
├── README.md           # from above
├── script.py
├── data.db
├── data/
│   ├── log1.txt
│   ├── log2.txt
│   └── ...
└── knowledge/
    ├── kubernetes.md   # declares K8s tools
    └── networking.md   # declares networking tools

Then update README.md with more tools and instructions:

---
tools:
  - [date]
  - [grep, -r]
  - [python3, script.py, { }]
  - [sqlite3, data.db, { regex: "^SELECT\\b.*" }]
---

Instructions
Check the current timestamp with date

Search through files with grep

Analyze and summarize logs with script.py

Run read-only queries against data.db

Browse ./knowledge for infrastructure context

5. Deploy it

Optionally, create a free Statespace account and deploy your app to the cloud:

statespace deploy . --public

More examples

See the examples/ directory for ready-to-run apps:

Concepts

Tools — Give agents controlled access to CLI commands over HTTP.

---
tools:
  - [date]
  - [grep, -r]
  - [python3, script.py, { }]
  - [sqlite3, data.db, { regex: "^SELECT\\b.*" }]
---

Components — Render live data inside pages with component code blocks.

Instructions — Guide agents through your data, workflows, and pages.

# Instructions
Check the current timestamp with date
Search through files with grep
Analyze and summarize logs with script.py
Run read-only queries against data.db
Browse ./knowledge for infrastructure context

Features

✅ Simple — It's just Markdown. Easy to learn, easy to use, easy to maintain.

⚡ Lightweight — Install a single, lightning-fast Rust binary. No dependencies.

🌐 Universal — Works directly with any agent that can make HTTP requests.

📦 Portable — Run or deploy your apps with a single CLI command.

🔒 Secure — Restrict access to your private apps with token-based authentication.

Community & Contributing