MiniMako is a security and optimization layer for AI coding agents like Claude Code, Codex, and Cursor. It monitors agent actions in real time, blocks dangerous commands before they execute, and maintains a clean, organized workspace by removing duplicate skills and optimizing context files. The platform provides an Agent Health Score and token analytics to help developers maximize performance while keeping their data private.
Funding
Funding not disclosed
Founders
Product
Problem
AI coding agents operate with broad permissions and can misinterpret instructions, leading to destructive actions like deleted databases, leaked credentials, or corrupted project files. Developers also struggle with cluttered agent configurations, conflicting plugins, and inefficient token usage, which degrades performance and increases costs.
Solution
MiniMako provides a governance and optimization layer for AI agents, giving them more autonomy while enforcing safety guardrails. The platform monitors every file access, command, and network connection in real time, blocking dangerous actions before they execute. It also maintains a structured "second brain" of files, rules, and memory to improve agent precision, while automatically cleaning up duplicate skills, conflicting plugins, and bloated context files. A daily dashboard and Agent Health Score show whether the setup is optimized, healthy, or broken, and token analytics reveal where spending is wasted.
Target Audience
Primary users are freelance developers, consultants, and creative professionals who rely daily on AI coding tools like Claude Code, Cursor, VS Code, and Codex for production work with real data and meaningful projects.
Features
- Real-time action monitoring with automatic blocking of dangerous commands, such as database deletions or unauthorized file access
- Detection and suspension of infected or malicious skills before they can propagate
- Automated workspace cleanup: removal of duplicate skills, flagging of conflicting plugins, and optimization of CLAUDE.md files
- Agent Health Score (0-100) that evaluates whether the agent ecosystem is optimized, healthy, or degraded
- Token and model-level analytics to identify wasteful spending and performance bottlenecks
- Privacy protection: tokens, passwords, and private files are prevented from leaking to the network