Automate iterative code optimization loops with structured JSON experiment results.
npx skills add https://github.com/yoonhwan/ai-feature-pack --skill auto-yoonhwan
Please help me install this Agent Skill. Skill: auto Source: https://github.com/yoonhwan/ai-feature-pack/tree/main/feature-pack/auto/skill Command: npx skills add https://github.com/yoonhwan/ai-feature-pack --skill auto-yoonhwan
Manual code optimization via trial and error requires constant manual test execution, result tracking, and change rollback, which is slow and prone to human error. This skill automates the entire experiment loop so agents can focus solely on code modifications while all infrastructure and evaluation tasks are handled automatically.
Use the auto skill to run a 2 hour autonomous experiment loop to optimize the character error rate of the local speech-to-text project.
High-intent search queries and answers about installing and using this skill.
Automating iterative code optimization requires an autonomous experiment loop that executes tests, parses structured JSON results, and applies code changes without manual intervention. This skill automates this workflow, handling test execution, result tracking, and change rollback automatically.
The Karpathy AutoResearch pattern enforces a strict agent role limited exclusively to code modifications, eliminating accidental infrastructure changes or manual test runs. This ensures the agent focuses solely on code changes while all evaluation tasks are handled automatically.
Tracking machine learning experiments automatically using git branches involves dedicating a branch to each experiment, executing the runner, and logging keep, discard, or crash verdicts to a TSV file. This skill handles this git-based experiment tracking automatically.
Running autonomous experiment loops requires a uv-managed Python environment, a project-specific experiment runner that outputs structured JSON results, and a goal declaration file. These prerequisites ensure the agent can execute and evaluate iterations continuously.
Running a long autonomous experiment loop for algorithm performance involves setting a predefined quantitative goal and allowing the agent to iterate continuously on code changes. This skill automates the entire loop, parsing results and rolling back changes until the goal is met.
Automated git-based experiment tracking eliminates the slow, error-prone manual tasks of executing tests, tracking results, and rolling back changes during code optimization. It allows agents to focus solely on code modifications while infrastructure and evaluation are handled automatically.
TeiNamEngineer autonomous loops with goal completion contracts and verification gates in Kiro.
ryan-mahoneyEvaluate GitHub issue specs for gaps, ambiguities, and feasibility concerns.
iamsteveholmesIdentify testing frameworks, runners, and configuration files across codebases.
pmcdowallTrack machine learning experiment metrics, configurations, and model artifacts with Weights & Biases.
