Danny Huang

Sentry is critical for Termdock and our backend services across error monitoring and performance tracing. Its issue grouping and release tracking let us map multi‑project, multi‑workspace events to the correct versions. Source Maps and Stack Traces speed up debugging, while Performance tracing visualizes slow transactions and bottlenecks. With alert rules and environment tags, we can quickly pinpoint issues in a terminal‑driven workflow and trace them back to the corresponding commit. We evaluated New Relic, Datadog, and Firebase Crashlytics. They excel at broad observability and mobile crash reporting, but we prioritize consistent frontend‑backend error context, fine‑grained release/environment segmentation, and a developer‑friendly debugging experience. As a result, we chose Sentry as the core to ensure stability, speed, and traceability when multiple projects run in parallel.

Termdock draws inspiration from vscode’s principles, but we intentionally took a different path. Termdock ships with multiple workspaces and a terminal‑driven workflow, supports Git isolation, and lets you operate several projects without interfering with each other. While Termdock is not a fork of vscode, vscode remains a highly professional and mature project—we learned a lot from its architecture and product design concepts. We evaluated Cursor, Zed, JetBrains Fleet, and several Web IDEs. They excel in editing experience and AI assistants, but the combination of native multi‑workspace support, deep terminal integration, and Git isolation didn’t fully meet our needs. Therefore, we chose the Termdock approach: terminal‑centric with workspaces as logical boundaries to ensure stability and speed when running multiple projects in parallel.

Claude Code is an excellent AI assistant for development. For Termdock, it delivers stable performance in long‑context understanding, structured refactoring, and controllable command generation—well suited to our terminal‑driven, multi‑workspace workflow. In practice, we use it to write and review scripts, generate test cases, summarize multi‑project changes into reviewable patches, and provide safe change suggestions for Git‑isolation scenarios. We evaluated Cursor, Windsurf, and Copilot / Copilot Chat. They offer strong inline completion and great in‑IDE experiences, but we prioritize long‑context reasoning, controllable command responses, and stability in terminal workflows. Therefore, we mainly chose Claude Code to align with Termdock’s approach: terminal‑centric with workspaces as logical boundaries to ensure stability and speed when running multiple projects in parallel.