ECC - An open-source, cross-platform AI agent performance optimization system

What is ECC?

ECC (Everything Claude Code) is a cross-platform AI Agent performance optimization system. The system provides 63 professional agents, 249 on-demand loaded skills, cross-session memory persistence, and the AgentShield security auditing tool for 7+ AI programming tools, including Claude Code, Codex, and Cursor. Through a Hook mechanism, it automates workflows, transforming AI programming assistants into well-organized AI teams with clearly defined roles.

Main functions of ECC

• 63 professional agentsIt covers roles across the entire development chain, including architecture design, code review, security audit, test-driven development, build bug fixing, and documentation updates.
• 249 On-Demand Loading SkillsSkills are dynamically matched based on the project's technology stack. TypeScript projects automatically load TS review skills, while Python test scenarios trigger TDD skills, avoiding context overload.
• Cross-session memory persistenceThe Hook mechanism automatically accesses the context between different sessions, enabling uninterrupted continuous workflow.
• AgentShield Security AuditIt includes 1282 built-in tests and 102 static analysis rules, providing millisecond-level scanning for credential leakage, misconfiguration, and injection risks.
• Continuous Learning v2Intuitive learning systems automatically extract reusable patterns from conversations, generate Instincts, and cluster them into Skills.
• Validation loops and checkpoints:support /checkpoint Save verification status/quality-gate Quality access control and /eval Evaluation framework.
• Multi-agent collaboration:support /multi-plan,/multi-execute Commands for multi-model collaborative workflows.
• Dashboard GUIIt provides the Tkinter desktop application, which supports theme switching, font customization, and visual browsing of components.

ECC Technical Principles

• On-demand loading architectureIt employs a Manifest-driven installation pipeline, stores state to track installed components, and injects Skills into the context only when the corresponding technology stack is detected.
• Hook event systemBased on the Claude Code v2.1+ Hook mechanism, it triggers automated scripts at 8 event points, including SessionStart, PreToolUse, PostToolUse, and Stop.
• Contextual Budget Management:pass ECC_SESSION_START_MAX_CHARS Environment variables control the upper limit of session startup context, and strategic compression is recommended to prevent token overflow.
• AgentShield Anti-Scan:--opus The mode initiates a three-agent adversarial pipeline consisting of a red team agent (finding vulnerabilities), a blue team agent (repairing vulnerabilities), and an auditor agent (summarizing and evaluating vulnerabilities).
• Selective installation architecture:support minimal,core,full Multiple installation configurations are available, and users can copy rules, agents, and skills as needed.
• Cross Harness DRY AdapterHook events on platforms like Cursor are accessed via... adapter.js Convert to Claude Code format and reuse the same script logic.

How to use ECC

• Plugin installation:
  • Execute in Claude Code /plugin marketplace add https://github.com/affaan-m/ECC Add market sources.
  • implement /plugin install ecc@ecc Install plugins to automatically load Skills, commands, and Hooks.
  • Manual copying required rules/ Directory to ~/.claude/rules/ecc/(The plugin does not support automatic rule distribution.)
• Manual installation:
  • Cloning repository git clone https://github.com/affaan-m/ECC.git.
  • Copy on demand agents/,skills/,commands/ arrive ~/.claude/ Corresponding directory.
  • run ./install.sh --profile minimal --target claude Install the Hook runtime.

ECC's core advantages

• Widest cross-platform coverageIt natively supports 7+ Harnesses, including Claude Code, Codex, Cursor, and OpenCod, allowing for reuse of multiple tools with a single configuration.
• Context-based intelligent management249 Skills are loaded on demand, and with the MCP tool's quantity control recommendations (no more than 10 MCPs per project), the 200K context window is effectively protected.
• Production-grade built-in securityAgentShield provides static analysis with 98% coverage and supports... --fix Automatic repair and --opus Deep confrontation audit.
• Self-evolution abilityThe continuous learning system extracts patterns from real development sessions, making the configuration increasingly tailored to individual workflows as it is used.

• Project official websitehttps://ecc.tools/
• GitHub repositoryhttps://github.com/affaan-m/ECC

ECC's Competitive Product Comparison

ECC application scenarios

• Full-stack project developmentA complete AI-assisted development pipeline, from requirements planning, architecture design, TDD coding to code review and E2E testing.
• Multi-technology stack teamIt can simultaneously maintain projects in multiple languages such as TypeScript, Python, Go, Java, Rust, Swift, and PHP, and automatically match the corresponding review agent.
• Security and compliance developmentAgentShield is used in financial and enterprise projects to prevent key leaks, misconfigurations, and injection attacks.
• Long-term maintenanceBy leveraging cross-session memory and continuous learning, AI can maintain its understanding of large codebases across multiple sessions.
• Hackathons and Rapid PrototypingDrawing on the author's experience in winning the championship, the entire chain from customer research to product prototype was built within a few hours.