Traction scoreGitHub stars can be faked, so popularity alone can be misleading. Traction Score looks for broader signs of recent attention, adoption, and active maintenance.
Codex Autoresearch manages the triage and execution of measured improvement loops within local projects. It is designed for tasks where success is defined by a concrete metric, such as reducing test runtime, improving build speed, or lowering query latency.
Performance engineers and ML researchers can use the tool to run repeated experiments against a specific target, such as model loss, while keeping non-measurable work like documentation and architecture direct.
WHO IT'S FOR
Performance engineers
running optimization loops for measurable targets
ML / Data scientists
optimizing model loss through repeated experiments
Frontend / Web performance engineers
improving Lighthouse scores and bundle sizes
AI agent / automation builders
integrating autonomous optimization loops into workflows
Compatible AgentsThe repository documents support for these agents. The skills may also work with other agents that can load SKILL.md files, but they may need some setup or small changes.
Open Codex in the local repository you want to improve.
The benchmark, correctness-check commands, and parser paths are author-provided examples; replace them with matching commands and paths in your project.
Provide a runnable benchmark and correctness checks with their dependencies already available.
Review and accept the proposed contract before setup or execution.
Use codex-autoresearch to propose a measured loop to reduce parser runtime.
Benchmark: node bench/parser.mjs
Metric: seconds (s), lower is better
Checks: node --test tests/parser.test.mjs
Scope: src/parser.mjs
Protect bench/parser.mjs and tests/parser.test.mjs.
Measure two baselines and repeat each candidate before a keep.
Stop after 5 packets or 30 minutes; repeat measurements count toward that limit.
Propose the complete contract for approval before setup or execution.
Codex