agnix
agnix is an open-source linter and LSP for developers using AI coding assistant configs, helping them detect and fix errors in files like CLAUDE.md, AGENTS.md, hooks, and MCP setup.
Tool overview
Based on the available evidence, agnix looks worth trying as an “AI coding assistant config validator,” but the adoption judgment is still early-stage rather than broadly proven. The heat proof is mostly a few X posts plus the GitHub repo description, which shows attention around the problem. The usability proof is thinner: mainly the official repo, author demo-style posts, and one user post saying it found three CLAUDE.md errors that were breaking agent context. That suggests real utility, but the sample is still limited.
Its practical role is not code generation, not a general-purpose IDE linter, and not an MCP server itself. A more accurate analogy is “ESLint/LSP for the instruction and wiring files around AI coding assistants.” The evidence supports validation for CLAUDE.md, AGENTS.md, SKILL.md, hooks, and MCP setups, with autofix plus IDE/LSP integration. That makes it useful for catching broken config, silent hook failures, or malformed assistant instructions before they cause odd agent behavior.
On cost and setup, the evidence only supports a few concrete points: it is open source, and the author says on X that it uses MIT/Apache-2.