release: v6.25.0 - Scientific Benchmarking Command

Add bashrs bench command for deterministic, scientifically rigorous shell script benchmarking.

NEW FEATURES:
- Scientific benchmarking with warmup + measured iterations
- Statistical analysis (mean, median, stddev, min, max, variance)
- Environment capture (CPU, RAM, OS, hostname)
- Quality gates integration (--strict linting, --verify-determinism)
- Multiple output formats (console, JSON)
- Comparison mode for multiple scripts
- Quiet mode for automation

IMPLEMENTATION:
- EXTREME TDD methodology (RED → GREEN → REFACTOR)
- 17 comprehensive CLI integration tests
- 8 unit tests for statistical functions
- NASA-quality specification (docs/specifications/bench-command.md)
- 550+ lines of implementation (rash/src/cli/bench.rs)

QUALITY METRICS:
- All 17 bench tests passing
- All 6000+ existing tests passing (zero regressions)
- 642 property tests passing (project-wide)
- Clippy clean, rustfmt formatted
- Real-world examples tested

DEPENDENCIES:
- Added sysinfo = "0.31" for system information detection

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <[email protected]>

Author: noahgift

CHANGELOG.md

@@ -7,6 +7,67 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [6.25.0] - 2025-11-01
+
+### ✨ NEW FEATURE - Scientific Benchmarking (EXTREME TDD)
+
+**bashrs v6.25.0 adds `bashrs bench` command for deterministic, scientifically rigorous shell script benchmarking.**
+
+### Added
+
+**`bashrs bench` Command** (17 comprehensive tests, 550+ lines, NASA-quality specification):
+
+- **Scientific benchmarking**: Warmup iterations + measured runs with statistical analysis
+- **Statistical metrics**: Mean, median, standard deviation, min, max, variance
+- **Environment capture**: CPU, RAM, OS, hostname metadata using sysinfo crate
+- **Quality gates integration**:
+  - `--strict`: Run bashrs linter on scripts before benchmarking
+  - `--verify-determinism`: Verify scripts produce identical output across runs
+- **Output formats**:
+  - Console: Beautiful formatted output with emojis and box-drawing characters
+  - JSON: Machine-readable results with full metadata (`--output results.json`)
+- **Comparison mode**: Benchmark multiple scripts, calculate speedup ratios, identify winner
+- **Quiet mode**: Suppress console output, only produce JSON (`--quiet`)
+- **Raw results**: Display all iteration times (`--show-raw`)
+
+**CLI Usage**:
+```bash
+# Basic benchmark
+bashrs bench script.sh
+
+# Custom iterations and warmup
+bashrs bench script.sh --iterations 20 --warmup 5
+
+# Compare multiple scripts
+bashrs bench fast.sh slow.sh
+
+# With quality gates
+bashrs bench script.sh --strict --verify-determinism
+
+# JSON output for automation
+bashrs bench script.sh --output results.json --quiet
+```
+
+**Specification**: Full NASA-quality spec at `docs/specifications/bench-command.md`
+
+**Implementation Methodology**:
+- EXTREME TDD: RED (17 failing tests) → GREEN (all passing) → REFACTOR (quality gates)
+- Property-based testing ready (infrastructure in place)
+- Test coverage: 17 CLI integration tests + 8 unit tests
+- Zero regressions: All 6000+ existing tests still passing
+
+**Quality Metrics**:
+- ✅ All 17 CLI tests passing
+- ✅ All 6000+ existing tests passing (no regressions)
+- ✅ Clippy clean
+- ✅ Formatted with rustfmt
+- ✅ Real-world examples tested
+- ✅ 642 property tests passing (project-wide)
+
+### Dependencies
+
+- Added `sysinfo = "0.31"` for CPU/RAM/OS detection
+
 ## [6.24.3] - 2025-11-01
 
 ### ⚡ PERFORMANCE - Code Complexity Reduction (EXTREME TDD)

Cargo.toml

@@ -29,7 +29,7 @@ rstest = "0.25"
 rustyline = "17.0"
 
 [workspace.package]
-version = "6.24.3"
+version = "6.25.0"
 edition = "2021"
 authors = ["Pragmatic AI Labs"]
 license = "MIT"