AI Config

Manage AI prompts
per environment.

Version and deploy AI agent instructions, skills, and rules as Markdown files — with per-environment overrides and MCP server integration.

Get started for free Read the docs

Write AI instructions in Markdown

A native editor for the format LLMs understand best. Organize agents, skills, and rules as files.

AI Config Editor

production saved

agent.md

code-review.md

testing.md

# Customer Support Agent

You are a customer support specialist for FlagDash. You help users with feature flags, remote config, and AI config management.

## Core Rules

- Always be concise and direct in responses

- Reference official documentation when possible

- Escalate billing questions to the support team

## Tone

Professional but friendly. Use technical terms when appropriate but explain them for newcomers.

agent — One per environment, the core identity

skill — Reusable capabilities and tasks

rule — Constraints and guardrails

Per-Environment

Different AI instructions for development, staging, and production. Test prompt changes safely before going live. No deploy needed to update behavior.

development verbose logging

staging new tone test

production stable

MCP Native

Built-in Model Context Protocol server with 27 tools. AI agents fetch their own instructions directly. Works with every MCP-compatible IDE and agent.

C

Claude Code connected

Cu

Cursor connected

W

Windsurf connected

Version Controlled

Every change to AI configs is tracked with full audit trail. Know who changed what prompt, when, and why. Roll back instantly if something breaks.

Updated agent tone

2 min ago

Added code-review skill

1 hour ago

Created initial agent.md

yesterday

API & MCP

Fetch AI configs via API.
Or let agents use MCP.

Read AI config files via REST API for your application, or connect AI agents directly through the built-in MCP server with 27 tools.

REST API for all AI config operations
Works with Claude Code, Cursor & Windsurf
Up to 50,000 characters per file

terminal

200 OK

Request

curl -H "Authorization: Bearer sk_your_api_key" \

https://flagdash.io/api/v1/ai-configs

Response

{

"ai_configs": [

{

"file_name": "agent.md",

"file_type": "agent",

"content": "# Assistant\nYou are..."

}

]

}

AI config management in three steps

From Markdown to production AI behavior in minutes.

1

Write Markdown

Create agent instructions, skills, and rules as .md files. The native format for LLMs.

# Agent

You are a helpful assistant.

Always respond in English.

2

Deploy per environment

Different prompts for development, staging, and production. Test freely without affecting live users.

development verbose mode

staging testing new tone

production stable release

3

Agents read via MCP

AI agents fetch their own instructions. Or your app reads configs via REST API. Always in sync.

MCP connected

use_ai_configs → loaded

get_ai_config → agent.md

Ship AI prompts like code

Manage agent instructions, skills, and rules as versioned Markdown files. Update AI behavior without redeploying.

Markdown-Based Files

Write AI config as Markdown — agent.md, skills, and rules. Native format for LLMs, human-readable for your team.

Per-Environment Overrides

Different AI instructions for development, staging, and production. Test prompt changes safely before going live.

MCP Server Integration

Built-in Model Context Protocol server. AI agents can read and update configs directly via MCP tools.

Agent, Skill & Rule Types

Three file types for structured AI management. One agent per environment, unlimited skills and rules.

Folder Organization

Organize AI configs into folders. Group related skills and rules together for cleaner management.

Versioning & Audit Trail

Every change to AI configs is tracked. Full audit trail of prompt modifications with timestamps and authors.

Ready to manage AI prompts at scale?

Version your AI configs per environment. Connect your agents via MCP. Set up in minutes.