Skip to content
MCP Server

Your AI agents connect directly.

57 tools and 4 resources over the Model Context Protocol. Let Claude Code, Cursor, and Windsurf manage your flags, configs, and AI prompts natively.

Discover tools✦ Draft safely✦ Approve changes✦ Release with agents✦

Works with the editors you already use

Any MCP-compatible agent can connect to FlagDash out of the box.

Claude Code
Cursor
Windsurf
Any MCP Agent
Live Demo

See how agents talk to FlagDash

A real conversation between an AI agent and the FlagDash MCP server. Natural language in, structured actions out.

MCP Session
Connected to FlagDash
AI Agent

Create a feature flag called new-checkout with type boolean.

FlagDash MCP
{
"status": "created",
"key": "new-checkout",
"flag_type": "boolean",
"enabled": false
}
next tool call
AI Agent

Toggle new-checkout to enabled and set the rollout to 25%.

FlagDash MCP
{
"status": "updated",
"key": "new-checkout",
"enabled": true,
"rollout_percentage": 25
}

57 tools. 4 resources. One endpoint.

Your AI agent can manage flags, configs, and AI configs without leaving the editor.

Feature Flags

10 tools

list_flags get_flag create_flag update_flag delete_flag toggle_flag set_flag_rollout set_flag_rules evaluate_flag evaluate_all_flags

Experiments

4 tools

list_experiments get_experiment create_experiment update_experiment

Session Replay

6 tools

list_session_replays get_session_replay list_experiment_replays get_session_replay_config update_session_replay_config delete_session_replay

Segments

5 tools

list_segments get_segment create_segment update_segment delete_segment

Remote Config

7 tools

list_configs get_config create_config update_config delete_config toggle_config set_config_value

Secrets

1 tool

manage_secrets

Translations

5 tools

list_translation_catalogs create_translation_namespace add_translation_locale put_translation_draft publish_translation_release

AI Configs

7 tools

use_ai_configs list_ai_configs get_ai_config create_ai_config update_ai_config delete_ai_config initialize_ai_configs

AI Releases

6 tools

list_ai_releases get_ai_release create_ai_release update_ai_release set_ai_release_rollout evaluate_ai_release

Skills

3 tools

list_skills get_skill get_skill_file

Projects & Environments

3 tools

list_projects get_project list_environments

Resources

4 auto-discovered resources

flagdash://ai-configs flagdash://configs flagdash://flags flagdash://projects
57
Tools
4
Resources
1
Config File
Setup

One config file.
Connected in seconds.

Add FlagDash to your MCP configuration and your AI agent instantly gains access to all 57 tools. No SDK installation, no dependencies.

.claude/mcp.json
JSON
{
  "mcpServers": {
    "flagdash": {
      "type": "sse",
      "url": "https://flagdash.io/mcp/sse",
      "headers": {
         "Authorization": "Bearer sk_your_api_key"
      }
    }
  }
}

Use an API key ( sk_ prefix) with read+write scopes for full access, or read-only scopes for evaluation only.

What your agent gets from FlagDash MCP

One config file, no SDK to install, and the same scoped keys you already use.

Zero SDK Setup

No packages to install, no dependencies to manage. The AI agent connects directly over MCP — just add one config file.

Auto-Discovery

Agents discover all available tools and resources automatically. No documentation lookup needed — the schema is the docs.

Full CRUD Access

Create, read, update, and delete flags, configs, and AI prompts. Anything you can do in the dashboard, your agent can do too.

use_ai_configs Tool

One tool call loads all agent instructions, skills, and rules. Your AI agent self-configures from FlagDash on every session.

Scoped API Keys

Give agents read-only or read-write access with scoped API keys. Principle of least privilege built in.

Open Standard

Built on the Model Context Protocol — an open spec from Anthropic. Any client that speaks MCP can connect.

Connect your AI agents today.

57 tools, 4 resources, one JSON config. Your agents manage flags, configs, and prompts from the editor.