📣 Hi All, 👋 I'm super excited about this one! 🚀Chimera v3.0 is back! And this is going to make your Claude (Desktop, Cloud and Code) x10 more powerful! A while ago, I decided to abandon the Chimera Project I had been working on for weeks. The reason was that sometimes you don't need to reinvent the wheel. As I was working through it and adapting it to the new Claude Subagents, I realized that there were much better versions of Chimera out there tackling the same problems and providing better solutions. There was no reason for me to pursue it any further, and I thought the best decision was to leave the project and use what was already available. BUT! Claude Code just released Skills this week, and this has brought Chimera back to life! 🤯 How? Let me explain what Skills are in my own words. 👇 As you know, all AI agents have system prompts where you can load instructions that the agent will follow every single time. They are very powerful, but they have their limitations. The problem with system prompts is that if you try to build a very complex system, you'll end up with a massive system prompt. Every single run will consume a vast amount of tokens, and your agent will quickly run out of context. This presents another problem: Agents are not able to use parts of the system prompt when they want; they have to run through the whole prompt every single time. A few months ago, we got something great that has changed the way agents call and use tools: MCPs. With MCPs, we were able to pick specific sets of tools so that our agents could use a whole bunch of third-party applications when needed, without having to define specific, single API calls. We could just request that our agent send an email, and as long as it was connected to a Gmail MCP, for example, the agent would find the right API endpoint and execute the action autonomously. The problem with MCPs is also a context problem. Every time the agents have to use MCPs, they run through the whole set of tools for each MCP and have to understand what every single call does before they can execute it. This consumes an immense number of tokens, depending on the complexity of the MCP server/client configuration.

Chimera architecture aligns remarkably well with what Anthropic has now built natively into Claude Code. Key Differences: What Chimera Has That Sub-Agents Don't (Yet) >Global persistent memory across projects via SQLite database >Cross-project learning and knowledge transfer >Automatic learning extraction and pattern synthesis >User preference tracking and adaptation over time >Complex orchestration with batch execution and parallel workflows What Sub-Agents Provide That Your System Lacks >Native integration with Claude Code's tool ecosystem >Automatic task delegation based on context matching >Separate context windows for each agent (prevents context pollution) >Built-in agent management via /agents command >MCP tool integration and granular permissions >Team sharing capabilities through version control My Migration Strategy: Hybrid Approach Rather than completely replacing Chimera, I will re-build a hybrid evolution that leverages the best of both approaches: Phase 1: Convert Specialist Agents to Sub-Agents format while preserving their memory capabilities. New File Structure: text .claude/├── backend.md # Native sub-agent format │ ├── designer.md # Native sub-agent format │ ├── datasci.md # Native sub-agent format │ ├── qa.md # Native sub-agent format │ ├── secure.md # Native sub-agent format │ ├── devops.md # Native sub-agent format │ └── ... (all specialists) ├── commands/ # Keep for orchestration │ ├── orchestrate.md │ ├── executor.md │ ├── learner.md │ └── memory_init.md Expected Benefits Immediate Gains: - Native tool integration and MCP compatibility - Automatic task delegation reduces manual orchestration - Separate context windows prevent context pollution - Built-in team sharing through .claude/agents/ versioning Long-term Advantages: - Hybrid intelligence combining native efficiency with learned wisdom - Cross-project knowledge transfer with modern delegation - Scalable architecture that grows with Claude Code's evolution - Future-proof design that adapts to new sub-agent capabilities

The Virtual Company Project: Finance Department The first block of the Virtual Company project will be the Finance Department. I've broken it down into four distinct functions to start, but there will be more—such as Financial Analyst and Payroll Specialist—in the future. Currently, I don't have enough data to simulate these additional roles. The four roles: - Victor (Controller):This workflow will receive bank transactions in the form of bank statements and transfer this information into the accounting books for future reconciliation with Accounts Payable (AP) and Accounts Receivable (AR). - Donna (Accountant #1):This workflow will process invoices dropped into a specific Google Drive folder as PDFs. The flow will be triggered, run OCR to extract all data into a structured JSON format, and then add the information into a simulated ERP using Google Sheets. - Mark (Accountant #2):Similar to the invoice process, but focused on expenses. I also aim to implement a more dynamic method for inputting expenses, such as taking pictures and sending them via email, WhatsApp, or Telegram to the workflow, which will process them and add them to the books. - Andrew (CFO):This AI agent will manage and extract all information related to financial data. It will be capable of pulling data from all books and responding to financial data requests made by the AI Orchestrator agent (CEO). This agent will interact with human users indirectly via the orchestrator. The workflows are currently at version 1. There are many improvements and additions that can be incorporated into the entire Virtual Company project for financial operations, but for now, I'm satisfied with the progress of this first block. This is just a concept, but with a few adjustments tailored to the specific software and frameworks each company uses, it can be easily adapted for real-world application—which is the ultimate goal of this project. Stay tuned for next chapter!

A few weeks ago I floated the idea of mapping an entire company structure with AI agents filling every role. I wasn’t sure it would fly, but the more I explored it, the more convinced I became: this is the future of truly agentic systems. 👍 Automating email replies? Cool. 👍 Auto-posting on social? Also cool. 👍 Following up with leads on autopilot? Super useful. 👍 AI agents answering phone calls? Huge time-saver. 🤯 The real magic happens when every one of these functions (and many more) connects seamlessly inside a business, so people can focus on what they do best: their craft. No more juggling calls, inboxes, invoices, or appointment confirmations. AI agents can already handle all of it. So... it begins! Meet Tony and his team, the virtual CEO and crew I’ve created to orchestrate this agentic operation. This post kicks off a new series where I’ll share the project’s progress. Stay tuned!