Speelka Agent
@korchasa
About Speelka Agent
Universal LLM Agent based on MCP
Overview
What is Speelka Agent?
Speelka Agent is a universal LLM agent based on the Model Context Protocol (MCP) with support for external tools, flexible configuration, and extensible logging. It runs as a Go binary on any platform supporting Go, and is aimed at developers building multi-agent workflows that orchestrate multiple MCP servers.
How to use Speelka Agent?
Clone the repository, build with go build ./cmd/server, then run the binary. Prepare a configuration file (YAML or JSON) or set environment variables (prefixed with SPL_). Use ./speelka-agent for stdio mode or ./speelka-agent --daemon [--config config.yaml] for HTTP mode. The agent acts as an MCP server itself, so any MCP client can connect to it.
Key features of Speelka Agent
- Multi-agent orchestration via external MCP server tools
- Flexible configuration: YAML, JSON, environment variables, overlay
- Extensible logging with multiple outputs (stdout, stderr, file, MCP)
- Security: key isolation, log protection, tool access control
- Built-in retry with exponential backoff for LLM and MCP errors
- Token and cost budgeting per request
Use cases of Speelka Agent
- Improve accuracy by splitting complex instructions into specialized tasks
- Reduce cost by using different LLM models for different task parts
- Extend, narrow, or modify responses from third‑party MCP servers
- Switch between real and LLM‑based tool implementations
- Enforce per‑request token and cost budgets for predictable usage
- Orchestrate multi‑step workflows across multiple MCP tools in one session
FAQ from Speelka Agent
Which LLM providers does Speelka Agent support?
OpenAI and Anthropic, configurable via the provider field (e.g., "openai", "anthropic") and model name (e.g., "gpt-4o", "claude-3-opus-20240229").
How can I configure Speelka Agent?
Configuration can be provided using YAML files, JSON files, or environment variables prefixed with SPL_. Example files are available in site/examples/.
Is Speelka Agent free and open source?
Yes, the source code is publicly available on GitHub at github.com/korchasa/speelka-agent-go under an open‑source license (details not specified in the README).
What transport protocols does the agent support?
The agent supports MCP over stdio (command line) and HTTP (daemon mode). You can enable or disable each transport in the configuration.
Does it work with any MCP server?
Yes, it connects to any MCP server defined in connections.mcpServers, supporting both command/args (for stdio servers) and configuration for HTTP servers.
Basic information
Category
Other
Author
More Other MCP clients
Roamzy
roamzy-ioAgent-native global eSIM over MCP — an AI agent buys one eSIM for 193 countries, anonymously, per-MB in USDT/USDC crypto, and earns 20% referral forever. Claude/Cursor/Continue + any MCP client.
tinyagents
albertvillanovaTiny Agents: LLM + MCP Tools
MCP Connect
EvalsOneEnables cloud-based AI services to access local Stdio based MCP servers via HTTP requests
MCPChatbot Example
keli-wenA chatbot implementation compatible with MCP (terminal / streamlit supported)
Osmosis-MCP-4B
gulp-aiAn Open Source SLM Trained for MCP
Comments