Iteratively propose edits, run experiments, and commit metric improvements.
npx skills add https://github.com/marlandoj/zouroboros --skill autoloop
Please help me install this Agent Skill. Skill: autoloop Source: https://github.com/marlandoj/zouroboros/tree/main/packages/workflow/docs/autoloop Command: npx skills add https://github.com/marlandoj/zouroboros --skill autoloop
Autoloop removes the manual, repetitive process of iterating code or configuration changes to improve a single numeric metric by automating propose→run→measure→commit loops and handling regressions automatically.
Start an autoloop campaign by running bun Skills/autoloop/scripts/autoloop.ts --program /path/to/your/project/program.md to iterate edits, run experiments, measure the metric, and persist improvements via git commits.
High-intent search queries and answers about installing and using this skill.
Automating iterative code edits to optimize a single numeric metric requires a continuous propose, run, and measure loop. This Skill autonomously proposes file edits, runs experiments, extracts metrics, and commits improvements via git while reverting regressions.
Yes, you can use an autonomous optimization loop for trading backtest experiments. It supports any workflow with a clear single numeric metric and a non-interactive run command, including prompt optimization and site performance tuning.
To start an autonomous metric optimization campaign, run the autoloop script with a path to your program.md file. This program.md must define your target file, run commands, and extract commands within a git repository.
Yes, you need a git repository to run autonomous experiment loops. The optimization process relies on git commits and branch management to persist metric improvements and automatically revert regressions.
Stagnation detection, retry attempts on crashes, and cost, duration, and experiment limits prevent runaway automated optimization loops. These guardrails ensure campaigns stop automatically when hitting constraints or failing to improve.
MCP integration for monitoring optimization campaigns exposes tools to start, monitor, list, stop, and fetch results. This enables direct integration with AI assistants and external services to manage campaign states.
BordaScan Python/ML codebases to generate program.md run specs with metric and guard configurations.
SakshamKandelAnalyze team progress from git commit history for weekly retrospectives.
hkqaiOrchestrate closed-loop autonomous optimization of experimental conditions for materials synthesis.
snoodleboot-ioExplain hyperparameter tuning techniques including grid search, random search, Bayesian optimization, and reinforcement learning.
