cctape: Local proxy to record Claude Code interactions.

The proliferation of local AI development, particularly around large code models like those powering Claude Code, has exposed a critical infrastructure gap: persistent, searchable context. Enter cctape, a dedicated local proxy designed to fill this void. In essence, cctape wraps the entire communication channel between the developer and the AI, acting as a transparent, yet powerful, recording mechanism. It intercepts every request and response, storing the complete session data in a local database. From a technical standpoint, this proxy approach is elegant. Instead of relying solely on the potentially truncated or context-limited API window, cctape establishes a single point of truth. This enables developers to perform full-text searches across months of sessions, find obscure decisions made in previous weeks, and even inspect the raw underlying API traffic. This level of observability is invaluable, especially when debugging complex, multi-stage AI workflows, making it an indispensable utility for professional AI engineering. Where cctape truly shines, however, is in its implementation of the MCP (Model Context Proxy) server. By registering itself with the Claude CLI, it empowers the AI agent to initiate searches against its own archive. This capability fundamentally shifts the cost-benefit analysis of AI usage; instead of the developer having to manually copy and paste necessary context back into the prompt—a costly and time-consuming process—the agent can locate and inject the relevant historical context itself. This significantly enhances the utility and autonomy of the AI for iterative development cycles. While the setup involves establishing a local HTTP service (`http://127.0.0.1:5555`), the developer experience is streamlined with clear setup guides for environment variables and VS Code extensions. The project structure suggests a modern, decoupled architecture (frontend/backend), indicating maintainability and extensibility. For any developer heavily invested in Claude Code or similar local LLM workflows, cctape moves beyond being a mere logging tool; it is a foundational piece of developer infrastructure that optimizes both cost and cognitive load.