Evidence-based prompt engineering. PromptLens tells you which phrases in your system prompt are actually doing work - and compresses the ones that aren't, with validation to prove nothing broke.

Local request viewer for Claude Code, Codex and OpenCode. Inspect prompts, tools, messages and responses in real time.Claude Code、Codex 和 OpenCode 的本地请求查看器。

Offline human prompt-pack debugger for testing agent expectations before LLM runtime.

Developer focused LLM experimentation and debugging server for testing prompts, tools, MCP integrations, traces, and scheduled AI workflows across models