Orchestrate iterative coding loops with worker model verification against success criteria.
npx skills add https://github.com/tomascupr/sous-chef --skill simmer-tomascupr
Please help me install this Agent Skill. Skill: simmer Source: https://github.com/tomascupr/sous-chef/tree/main/skills/simmer Command: npx skills add https://github.com/tomascupr/sous-chef --skill simmer-tomascupr
This skill solves the problem of iterative, goal-based development where a task requires multiple attempts to reach a specific, verifiable state, preventing the need for manual oversight of every single iteration.
Ask the assistant to simmer the current task by defining the goal, the verification commands, and the maximum number of laps to run.
High-intent search queries and answers about installing and using this skill.
Automated iterative coding loops delegate implementation to a worker model and verify results against defined success criteria, running multiple attempts until tests reach a stable passing state.
Goal-oriented development executes a series of implementation laps using a worker model, with a judge verifying progress against machine-checkable goals to prevent the need for manual oversight of every iteration.
Yes, automated coding loops enforce strict budget caps, no-progress detection, and branch-based development to ensure failed attempts do not corrupt the main codebase during complex regression fixes.
Automated code iteration executes implementation laps using Codex as the worker model, with Claude acting as the judge to verify progress against machine-checkable goals for trial and error development.
The best way to run multiple implementation laps is to define the goal, verification commands, and maximum number of laps, allowing the loop to iterate until the new feature reaches a verifiable passing state.
Automated coding loops are limited by strict budget caps and no-progress detection, stopping iteration if the maximum number of laps is reached or if failed attempts fail to show measurable progress.
daffy0208Convert high-level goals into executable HTN plans with capability matching and precondition validation.
obraRun tests, present four Git completion options, and execute the chosen workflow.
reomgitCompute a weighted composite code quality score from project tools.
aresbitRun non-interactive Gemini CLI prompts for Q&A, summaries, and generation.
