Coding & Development
BENCHMARKS.md Β· 7 documents
agentmark β Benchmark AI Coding Agents on Your Codebase
Defines an open-source Python CLI that benchmarks AI coding agents on a user's own codebase and tasks, producing a terminal comparison report of pass/fail, time, cost, tokens, and LLM calls.
πΊοΈ HeySeen Development Plan
Plans a multi-phase pipeline converting PDFs to LaTeX and images on macOS Apple Silicon, with completed milestones and next steps.
What If You Could Run 20 AI Agents in One Terminal?
Describes a prototype that runs multiple CLI coding agents in parallel tmux panes, each with its own workspace and task queue.
Development notes
Documents iterative model experiments for a financial returns prediction challenge, tracking what worked and what didn't across three versions.
Benchmark: pinky
Presents cycle-accurate NES emulator and prime sieve benchmarks comparing PolkaVM against 15+ other VMs across oneshot, execution, and compilation time.
Benchmarks
Documents how to run, compare, and interpret Criterion benchmarks for a Rust LSP project's parsers, caches, and version utilities.
Ruby 2.7
Documents every language, core class, and standard library change in Ruby 2.7 with code examples and rationale.