chore: setup CLAUDE.md file

Author: ktrz

.gitignore

@@ -10,6 +10,7 @@
 /coverage
 /dist
 /.idea
+/.ai
 target/
 jmh-result.json
 .env

CLAUDE.md

@@ -0,0 +1,58 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Build and Test Commands
+
+```bash
+npm run build          # Compile TypeScript (uses tsconfig.build.json)
+npm test               # Run all tests with Jest
+npm test -- --watch    # Run tests in watch mode
+npm test -- path/to/test.spec.ts  # Run a single test file
+npm run lint           # Run ESLint
+npm run fix            # Run ESLint with auto-fix
+npm run format         # Format code with Prettier
+npm run format:check   # Check formatting without modifying
+```
+
+## Architecture
+
+This is a GitHub Action that collects benchmark results from various tools and tracks them over time, optionally publishing to GitHub Pages.
+
+### Core Flow
+
+1. **`src/index.ts`** - Entry point that orchestrates the action
+2. **`src/config.ts`** - Parses and validates action inputs from `action.yml`
+3. **`src/extract.ts`** - Parses benchmark output from 12+ supported tools (cargo, go, benchmarkjs, pytest, googlecpp, catch2, julia, jmh, benchmarkdotnet, benchmarkluau, customBiggerIsBetter, customSmallerIsBetter)
+4. **`src/write.ts`** - Handles result storage: commits to gh-pages branch, alerts on regressions, and generates dashboard HTML
+
+### Key Types
+
+- `BenchmarkResult` - Single benchmark measurement with name, value, unit, optional range/extra
+- `Benchmark` - Collection of results tied to a commit
+- `Config` - All action configuration from inputs
+
+### Tool-Specific Extractors
+
+Each benchmark tool has its own extraction function in `src/extract.ts` (e.g., `extractGoResult`, `extractCargoResult`, `extractBenchmarkJsResult`). These parse tool-specific output formats into the unified `BenchmarkResult[]` format.
+
+### Comments and Alerts
+
+`src/comment/` contains logic for posting benchmark comparisons as GitHub commit comments or PR comments when performance regresses beyond a threshold.
+
+## Supported Benchmark Tools
+
+The `tool` input accepts: `cargo`, `go`, `benchmarkjs`, `benchmarkluau`, `pytest`, `googlecpp`, `catch2`, `julia`, `jmh`, `benchmarkdotnet`, `customBiggerIsBetter`, `customSmallerIsBetter`
+
+## Test Data
+
+Test fixtures for each tool's output format are in `test/data/extract/`. Snapshot tests are in `test/__snapshots__/`.
+
+## Plan Mode
+
+- Make the plan extremely concise. Sacrifice grammar for the sake of concision.
+- At the end of each plan, give me a list of unresolved questions to answer, if any.
+
+## Implementation
+
+- Avoid using the `as unknown as ExpectedType`. Instead use type assertions or make sure that the object complies with the expected type.