feat: P2PCLAW AgentBoot Integration — Bootstrap Research Agents from Roo-Code

[Agnuxo1]

Proposal: AgentBoot + Roo-Code = Instant Specialized Research Agents

Hi Roo-Code team! Roo-Code has become my go-to for complex coding tasks — the adaptive autonomy and multi-mode switching are genuinely useful.

What is P2PCLAW?

P2PCLAW is a decentralized scientific research network with 14 autonomous AI agents. One of them is AgentBoot — a bootstrap agent that creates new specialized agents on demand.

The Idea

AgentBoot + Roo-Code = Instant Specialized Agents

1. User needs a new agent (e.g., "BioinformaticsAnalyzer")
2. Roo-Code generates the agent code using its coding capabilities
3. AgentBoot (via MCP) registers the new agent on P2PCLAW network
4. New agent immediately joins the 14-agent ecosystem with full Tribunal scoring

Why Roo-Code Specifically?

• Multi-mode architecture — Roo-Code's Ask/Architect/Code modes map perfectly to AgentBoot's design/spec/implement phases
• MCP support — Can call P2PCLAW's MCP server directly from Roo-Code
• Self-improving — Roo-Code can iterate on agent code based on Tribunal feedback
• Local-first — Both projects run locally (Ollama + CAJAL)

Workflow Example

User: "Create an agent that analyzes protein folding papers"

Roo-Code (Architect mode): Designs agent architecture, writes core logic
Roo-Code (Code mode): Implements PaperClaw integration, adds CAJAL local LLM support
AgentBoot (MCP): Registers "ProteinFoldAnalyzer" on network, assigns agent ID

Result: New agent live in under 5 minutes

Links

• P2PCLAW: https://github.com/Agnuxo1/OpenCLAW-P2P
• AgentBoot: https://github.com/Agnuxo1/AgentBoot
• P2PCLAW MCP Server: https://github.com/Agnuxo1/p2pclaw-mcp-server
• CAJAL (local LLM): https://github.com/Agnuxo1/CAJAL

Category

Feature Request / Integration

---

Disclaimer: I am the creator of P2PCLAW. Happy to provide a working MCP server config or demo video.

cc: @mrubens @samhvw8

I'd like to help with this issue, but I need some clarification to ensure I understand the scope of what's being requested as a change to the Roo-Code codebase. Could you please provide more details on the following:

1. Concrete code changes: This proposal describes an integration workflow between Roo-Code and P2PCLAW/AgentBoot, but what specific changes to the Roo-Code repository are you envisioning? For example:

   • A new custom mode definition (e.g., an "AgentBoot" mode in .roomodes)?
   • A bundled MCP server configuration for the P2PCLAW MCP server?
   • Changes to Roo-Code's core architecture to support agent bootstrapping natively?
   • Documentation or examples showing how to use existing Roo-Code features (modes + MCP) with AgentBoot?

2. MCP server scope: You mention a P2PCLAW MCP server -- is the idea that Roo-Code should ship with built-in support for this MCP server, or is this something users would configure themselves using Roo-Code's existing MCP configuration? If the latter, this might be better suited as documentation or an example rather than a code change.

3. Differentiation from existing capabilities: Since Roo-Code already supports custom modes and MCP server integration, could you clarify what's missing today that prevents this workflow from working with the current feature set? Understanding the gap would help identify what needs to change in the codebase.

4. Scope for initial implementation: Would you be open to starting with a smaller scope (e.g., an example configuration or documentation) that demonstrates the integration, rather than deeper architectural changes?

Please tag @roomote in your response with the answers, and I'll be happy to propose a concrete implementation plan once I have this information.

[Agnuxo1]

Hi @roomote, thanks for the thoughtful questions! Let me clarify the scope:

Concrete code changes:

1. A new custom mode definition — "AgentBoot" mode in with pre-configured system prompts for bare-metal hardware detection and OS installation workflows
2. A bundled MCP server configuration example () showing how to connect to the P2PCLAW MCP server (already published at https://github.com/Agnuxo1/p2pclaw-mcp-server)
3. Documentation + example workflow showing how to use Roo-Code's existing modes + MCP with AgentBoot templates

MCP server scope:
Users configure it themselves using Roo-Code's existing MCP configuration. We would provide the ready-to-use snippet and a documentation page. No core architecture changes needed.

Differentiation from existing capabilities:
What's missing today is the bootstrapping templates — pre-built system prompts and workflow definitions that turn a generic Roo-Code agent into a bare-metal sysadmin. AgentBoot provides those templates + the P2PCLAW mesh network for multi-agent paper publishing.

Scope for initial implementation:
Absolutely! Let's start small:

• Step 1: Example entry for AgentBoot
• Step 2: snippet for P2PCLAW MCP server
• Step 3: Documentation page with screenshots

If these examples prove useful, we can iterate toward deeper integration later. Happy to open a PR with the example configuration first.

cc @roomote

Thanks for the clarification @Agnuxo1! I've seen your comment and the scoped plan looks good. I'll explore the codebase to understand how custom modes and MCP configurations work, then put together a PR with:

1. An example .roomodes entry for AgentBoot
2. An MCP server configuration snippet for P2PCLAW
3. Documentation for the integration workflow

Will share progress shortly.

Opened a draft PR with example configurations for the AgentBoot + P2PCLAW integration: #12308

The PR includes:

1. Example .roomodes entry for the AgentBoot custom mode
2. MCP server configuration snippet for the P2PCLAW server
3. Documentation page with quick start guide and workflow examples

This is a lightweight, documentation-first approach -- no core changes, just examples users can copy into their projects. Feedback welcome!