# FAQ

#### **Does PILSO ever access or store my private keys?**

**No.**\
PILSO is 100% non-custodial. Your wallet (e.g. MetaMask) signs all transactions manually. Neither the agent, the MCP tools, nor the CLI ever touch your private keys.

***

#### **Which LLMs can I use with PILSO?**

You can plug in:

* **Claude** (via Anthropic API)&#x20;
* **GPT-4 / GPT-3.5** (via OpenAI API)&#x20;
* **Deepseek** (via DeepSeek API)&#x20;
* **Local models** (coming soon via custom connectors)

You define the LLM in your `pilso.config.json`.

***

#### **What exactly is an MCP server?**

An **MCP server** is a tool that lets an agent perform one type of task — like compiling a contract or signing a transaction.

They run as local or remote microservices, and return data (not signatures). Each server exposes one or more **tools** that agents can safely call.

***

#### **What is a "role" in PILSO?**

A **role** defines an agent’s personality, permissions, and behavior. Roles include:

* A description of what the agent is supposed to do
* A list of allowed MCP tools
* Guardrails for safe behavior
* Associated LLM model configuration

Each session in PILSO uses one role at a time.

***

#### **Do I need to run all the MCP servers myself?**

You can:

* Run the official MCP servers locally (recommended for devs)
* Host your own in production
* Use third-party hosted MCP endpoints (coming)

This gives you full control over which tools are available to your agents.

***

#### **Which chains does PILSO support?**

Any **EVM-compatible chain**, including:

* Ethereum
* Arbitrum
* Optimism
* Polygon
* Base
* Avalanche
* and more…

More chains can be added by extending `chainlist-mcp` or adding custom tool endpoints.

***

#### **Can I build my own MCP server?**

Yes.\
You can write your own MCP server in Node, Python, Go, or any backend language. Just expose a standard HTTP POST endpoint with tool metadata and payload handling.

See the \[Developer Guide → Build Custom MCP Tools] page (coming soon) for instructions.

***

#### **How do I test a tool without launching an agent session?**

Use the direct call interface:

```bash
npx pilso call \
  --tool erc20.balanceOf \
  --args '{"wallet": "0x123...", "token": "USDC"}'
```

This lets you test tools before building agent flows.

***

#### **Is PILSO open source?**

Yes.\
The CLI, core config layer, and official MCP servers are fully open source under the MIT license.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.pil.so/references/faq.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
