Installation

ACENLY Bench requires Python 3.9+ and PyYAML. Download the two files from GitHub and drop them into your project root:

Or download directly without git:

★ View on GitHub →

Configuration

Create acenly.yml in your project root. Copy from acenly.example.yml as a starting point:

benchmark:

  track:
    - file: src/api.py
      function: deduplicate_users
    - file: src/search.py
      function: build_index

  # warn if 10% slower, block push if 25% slower
  regression_warn:  0.10
  regression_block: 0.25
  noise_floor:      0.05

  trials: 5
  warmup: 2

Config options

• track — list of functions to benchmark, each with file and function
• regression_warn — warn but allow push if this much slower (e.g. 0.10 = 10%)
• regression_block — block push entirely if this much slower (e.g. 0.25 = 25%)
• noise_floor — ignore changes smaller than this (filters measurement noise)
• trials / warmup — number of timed runs and warmup runs per function

Running

Benchmark all tracked functions:

Benchmark one specific function:

View history across commits:

High-precision mode — batch timing, 3-second measurement window:

Git Hooks

Install a pre-push hook that runs benchmarks automatically before every push:

If a function regresses past regression_block, the push is stopped. To override once:

Remove the hook:

Optimizer — Coming Soon

Once the benchmark finds something slow, the ACENLY Optimizer rewrites it algorithmically — O(n²) to O(n), correctness-verified, pure Python output. No new dependencies.

The optimizer is currently in private beta. We're onboarding a small group early.