A code reader that admits agents made it obsolete

What it does

CodeAsk is an Electron desktop app that pipes codebases into large language models and returns structured audits, security checks, and architectural summaries. It displays results in a split-pane viewer with Markdown rendering, Mermaid diagrams, and read-only code previews. A companion JetBrains plugin can load the resulting .codeaskdata reports directly inside the IDE.

The interesting bit

The README opens with a blunt admission that the project “no longer has any reason to exist” because modern agent tools have absorbed its use case. That level of self-awareness is rare in open-source marketing. Under the hood, the app serializes analysis results into shareable .codeaskdata files, letting teams swap LLM-generated code reviews without burning API tokens on re-runs.

Key highlights

• Desktop Electron wrapper around LLM-driven code analysis, built with React 19 and Zustand.
• Configurable plugin system for custom prompts, supporting both single-file and global codebase scans.
• Portable .codeaskdata output that can be opened by teammates without re-analyzing the repository.
• JetBrains IDE plugin (currently manual-install only from a separate repo) for viewing reports next to source.
• Explicit recommendation to use local Ollama deployments for confidential projects.

Caveats

• The author notes that most prompt templates were generated by DeepSeek and are largely untested, so expect to tune them yourself.
• The JetBrains plugin has not been published to the marketplace yet.
• A VS Code plugin is on the roadmap but still under construction.

Verdict

If you are stuck maintaining an undocumented legacy monolith and need offline, shareable analysis artifacts, this might earn a spot in your toolkit. Otherwise, the author has already pointed you toward the agent-based alternatives that rendered it redundant.