Expose your home PC. Develop from everywhere. Claude Code tabs in the browser. A deep inspector that shows every LLM interaction, tool calls, hooks, and subagent parallelism. Ctrl-V paste and file drop into CLI sessions. Route to any LLM API.
Features: Home PC Automation, Claude Code Tabs, Deep LLM Inspector, Paste & Drop, Provider Routing (OpenAI, Gemini, DeepSeek, Kimi, Ollama), Rules, MCP Server, File Manager.
Collaborative media collections fed by a Telegram/WhatsApp bot — solo or shared, with live per-reader translation. An AI builder turns plain-language descriptions into reusable layers — magazine, ad page, film — over the same shared timeline.
Strength-training tracker. Log workouts via a Telegram bot (text or voice), review progress on a 3D anatomical body map, get AI-powered readiness pushes.
I intercepted 307 files from a Claude Code session to reverse-engineer its system prompt, tool definitions, subagent architecture, and caching strategy. Tool schemas alone consume 56% of input tokens, prompt caching saves 90% of costs, and about 19% of the tool payload could be trimmed.
I intercepted every API call Claude Code makes and compared two sessions: a simple greeting (8 calls, 22 output tokens) vs. a full security review (81 calls, multi-agent orchestration). The same 19K-token system prompt powers both. Here's what I found about routing, caching, subagents, and the real cost of AI-powered development tools.
An analysis of Claude Code's two-layer planning system — plan mode and the Plan subagent — based on intercepted API calls, system instructions, and real session data from my inspector tool.
