D6E D6E
For developers

Everything is
a public API.

The D6E console is just a client of the same REST API you can call with an API key. The instance MCP server exposes the same ~90 tools the built-in chat agent uses. Anything the console can do, your code can do.

Read the developer docs

Design philosophy

D6E is designed to stay loosely coupled to the systems you already own: it hosts your work — it does not absorb your system.

Data and AI in your environment

The instance runs on your infrastructure; workspace data is never sent outside. The central site only brokers authentication.

Everything is a public API

The console and the built-in agent are built on the public REST API and MCP tools. There are no privileged internal APIs.

Portable assets, loose coupling

JS STFs are pure functions; Docker STFs are ordinary containers speaking JSON over stdin/stdout; prompts are Markdown; the database is plain PostgreSQL. D6E itself is the glue — auth, policies, audit, orchestration. If you ever migrate away, you rewrite the glue, not your logic.

Adopt incrementally

Start with the console, harden recurring procedures into STFs and workflows, package them as a Plugin, and add a custom frontend when a task deserves one. Stop at any stage and you still have a working system.

Governance at the boundary

Deny-by-default table policies, audit logs, usage tracking, and encrypted credential storage. Every consumer goes through the same API, so controls are enforced in one place.

Read the design philosophy

Feeding it to an AI agent? Fetch the raw Markdown directly:

https://www.d6e.ai/docs/design-philosophy.md

Your coding agent is the dev environment

Connect Codex CLI, Claude Code, Cursor, or any local AI agent to the instance MCP server and develop against the real workspace with the same tools the built-in agent uses. There is no separate SDK to learn.

Local AI agent development guide
.cursor/mcp.json
{
  "mcpServers": {
    "d6e": {
      "url": "http://<instance-host>:8081/mcp",
      "headers": { "Authorization": "Bearer d6e_YOUR_API_KEY" }
    }
  }
}

Install Agent Skills (recommended: all of them)

For design, Q&A, and custom frontend work, install Plugin, Docker STF, and custom-frontend skills together. A single-repo install makes it easy to misunderstand the boundary.

For Cursor / Claude Code / Codex. Run from your project root.

How to install Skills Read the developer docs
terminal
# Recommended: install all skills
npx skills add d6e-ai/d6e-plugin-skills --skill '*' -y
npx skills add d6e-ai/d6e-docker-stf-skills --skill '*' -y
npx skills add d6e-ai/d6e-custom-frontend-skills --skill '*' -y

What you can build

Each skills repository is a development guide your AI coding agent can consume directly. For design, Q&A, and custom FEs, prefer the all-skills install above.

Plugins

Package prompts, files, STFs, and workflows into a template.yaml, install from a URL, and optionally distribute on the Marketplace.

View the skills repository

Docker STFs

Run any language in a container. The contract is JSON in on stdin, JSON out on stdout — secrets are injected at run time.

View the skills repository

Custom frontends

An independent, ordinary web app that signs users in with OAuth2 and calls the public API. Usually combined with Plugins and Docker STFs.

View the skills repository

Start building on D6E

Create an API key in the console and connect your agent. Questions about deployment? We're happy to help.

Read the developer docs Contact us Local AI agent development guide