Sequential thinking via a multi-agent system built with Agno (planner, researcher, analyzer, critic, synthesizer) for structured reasoning.
Setup
Choose your client, copy the config, then fill in any required environment variables.
claude mcp add multi-agent-sequential-thinking --env DEEPSEEK_API_KEY='<DEEPSEEK_API_KEY>' -- uvx mcp-server-mas-sequential-thinking~/.codex/config.toml
[mcp_servers.multi-agent-sequential-thinking]
command = "uvx"
args = ["mcp-server-mas-sequential-thinking"]
env = { DEEPSEEK_API_KEY = "<DEEPSEEK_API_KEY>" }
# codex mcp add multi-agent-sequential-thinking --env DEEPSEEK_API_KEY='<DEEPSEEK_API_KEY>' -- uvx mcp-server-mas-sequential-thinking~/.cursor/mcp.json
{
"mcpServers": {
"multi-agent-sequential-thinking": {
"command": "uvx",
"args": [
"mcp-server-mas-sequential-thinking"
],
"env": {
"DEEPSEEK_API_KEY": "<DEEPSEEK_API_KEY>"
}
}
}
}.vscode/mcp.json
{
"servers": {
"multi-agent-sequential-thinking": {
"type": "stdio",
"command": "uvx",
"args": [
"mcp-server-mas-sequential-thinking"
],
"env": {
"DEEPSEEK_API_KEY": "<DEEPSEEK_API_KEY>"
}
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"multi-agent-sequential-thinking": {
"command": "uvx",
"args": [
"mcp-server-mas-sequential-thinking"
],
"env": {
"DEEPSEEK_API_KEY": "<DEEPSEEK_API_KEY>"
}
}
}
}claude_desktop_config.json
{
"mcpServers": {
"multi-agent-sequential-thinking": {
"command": "uvx",
"args": [
"mcp-server-mas-sequential-thinking"
],
"env": {
"DEEPSEEK_API_KEY": "<DEEPSEEK_API_KEY>"
}
}
}
}.trae/mcp.json
{
"mcpServers": {
"multi-agent-sequential-thinking": {
"command": "uvx",
"args": [
"mcp-server-mas-sequential-thinking"
],
"env": {
"DEEPSEEK_API_KEY": "<DEEPSEEK_API_KEY>"
}
}
}
}Environment variables
DEEPSEEK_API_KEY
Replace the placeholders with your own credentials and keep them out of version control.