Hica
Highly customizable Agent Library
What is Hica?
Hica is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to highly customizable agent library
Highly customizable Agent Library
This server falls under the Coding Agents category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- Highly customizable Agent Library
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx hicaConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use Hica
Hica is a highly customizable Python agent library built around a deterministic Thought-Action-Observation loop, offering full visibility into agent decision-making via structured event logging, native MCP tool integration alongside local Python functions, human-in-the-loop clarification workflows, and pluggable persistence backends for resumable conversation threads.
Prerequisites
- Python 3.12+ installed
- pip or uv package manager
- An LLM provider API key (e.g. OpenAI, Anthropic, or any LiteLLM-compatible provider)
- Optional: MongoDB, SQLite, or a file system path for persistent thread storage
- Optional: An MCP server to connect external tools
Install Hica
Install the hica package from PyPI. Use the [all] extra to include all optional dependencies for memory backends and MCP client support.
# Core install
pip install hica
# With all optional dependencies
pip install 'hica[all]'Define Tools with the Registry Decorator
Tools are registered with a simple decorator. Each tool function receives typed arguments and returns a result. Hica separates tool execution from LLM conversation management automatically.
from hica import registry
@registry.tool()
def add(a: int, b: int) -> int:
"Add two numbers together"
return a + b
@registry.tool()
def search_files(query: str, directory: str = ".") -> list:
"Search files matching a query in a directory"
import glob
return glob.glob(f"{directory}/**/*{query}*", recursive=True)Configure and Instantiate an Agent
Create an AgentConfig specifying your LLM provider and model using LiteLLM-style model strings. Pass the config and your tool registry to instantiate the agent.
from hica import Agent, AgentConfig
config = AgentConfig(
model="openai/gpt-4.1-mini", # or "anthropic/claude-3-5-sonnet-20241022"
)
agent = Agent(config=config, tools=registry)Run an Agent with Persistent Thread Storage
Threads store the full conversation and event log. Choose a persistence backend — file, SQL, MongoDB, or in-memory. Threads are resumable: pass the same thread ID to continue an interrupted session.
from hica import Agent, AgentConfig, FileThreadStorage
storage = FileThreadStorage(path="~/.hica/threads")
agent = Agent(config=config, tools=registry, storage=storage)
# Start a new thread
result = agent.run("Search for Python files in the current directory", thread_id="session-1")
# Resume the same thread later
result = agent.run("Now count how many you found", thread_id="session-1")Connect an MCP Server as a Tool Source
Hica supports MCP tool integration alongside local Python tools. Pass an MCP server command to the agent to expose its tools automatically inside the agent loop.
from hica import Agent, AgentConfig, MCPToolSource
mcp_source = MCPToolSource(
command="npx",
args=["-y", "@modelcontextprotocol/server-filesystem", "/tmp"]
)
agent = Agent(config=config, tools=registry, mcp_sources=[mcp_source])Configure Claude Desktop to Use Hica as an MCP Server
Hica can itself be exposed as an MCP server so Claude Desktop can invoke your custom agent and its tools. Add the following to your Claude Desktop config.
{
"mcpServers": {
"hica-agent": {
"command": "python",
"args": ["-m", "hica.server"],
"env": {
"OPENAI_API_KEY": "your_openai_key"
}
}
}
}Hica Examples
Client configuration
Claude Desktop config to expose a Hica agent as an MCP server with OpenAI as the LLM backend.
{
"mcpServers": {
"hica-agent": {
"command": "python",
"args": ["-m", "hica.server"],
"env": {
"OPENAI_API_KEY": "sk-..."
}
}
}
}Prompts to try
Example tasks that demonstrate Hica's deterministic agent loop, human-in-the-loop, and tool use.
- "Search for all Python files in /tmp and list their sizes."
- "Add 157 and 843 using the calculator tool."
- "I need you to ask me for confirmation before deleting any file."
- "Resume our previous conversation about the file search and summarize what we found."
- "Run the agent with both local tools and the filesystem MCP server, then list available tools."Troubleshooting Hica
ImportError: Python 3.12+ is required but an older version is installed
Hica requires Python 3.12 or newer. Check your version with 'python --version'. Use pyenv or conda to install Python 3.12: 'pyenv install 3.12.0 && pyenv local 3.12.0'.
Agent loop does not call the registered tool, even when the task is clear
Ensure the tool docstring clearly describes what the tool does — Hica passes docstrings to the LLM as tool descriptions. Also confirm the tool is registered before the Agent is instantiated, not after.
MCP tool source fails to connect with 'command not found'
The command specified in MCPToolSource must be on your system PATH. Test it manually (e.g. 'npx -y @modelcontextprotocol/server-filesystem /tmp') before passing it to Hica. Ensure npx or uvx is installed and accessible.
Frequently Asked Questions about Hica
What is Hica?
Hica is a Model Context Protocol (MCP) server that highly customizable agent library It connects AI assistants to external tools and data sources through a standardized interface.
How do I install Hica?
Follow the installation instructions on the Hica GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with Hica?
Hica works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is Hica free to use?
Yes, Hica is open source and available under the Apache-2.0 license. You can use it freely in both personal and commercial projects.
Hica Alternatives — Similar Coding Agents Servers
Looking for alternatives to Hica? Here are other popular coding agents servers you can use with Claude, Cursor, and VS Code.
Dify
★ 142.2kProduction-ready platform for agentic workflow development.
Ruflo
★ 54.0k🌊 The leading agent orchestration platform for Claude. Deploy intelligent multi-agent swarms, coordinate autonomous workflows, and build conversational AI systems. Features enterprise-grade architecture, self-learning swarm intelligence, RAG integrat
Goose
★ 45.7kan open source, extensible AI agent that goes beyond code suggestions - install, execute, edit, and test with any LLM
Antigravity Awesome Skills
★ 38.3kInstallable GitHub library of 1,400+ agentic skills for Claude Code, Cursor, Codex CLI, Gemini CLI, Antigravity, and more. Includes installer CLI, bundles, workflows, and official/community skill collections.
AgentScope
★ 25.5kBuild and run agents you can see, understand and trust.
Serena
★ 24.5kA coding agent toolkit that provides IDE-like semantic code retrieval and editing tools, enabling LLMs to efficiently navigate and modify codebases using symbol-level operations instead of basic file reading and string replacements.
Browse More Coding Agents MCP Servers
Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up Hica in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use Hica?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.