autonomous-loop

Automated iterative task execution applies code changes and runs checks repeatedly until success criteria are met or a maximum iteration limit is reached. It orchestrates development cycles hands-off, escalating to human triage only when necessary.

Jidoka escalation is a safety mechanism that halts automated iteration and triggers human triage when automated changes become unsafe or continuously fail success criteria. It prevents unsafe modifications by escalating through defined tiers.

Configuring iteration controls involves setting max_iterations to limit execution cycles and defining success_criteria to determine when checks pass. The loop logs each iteration and handles escalation automatically based on these parameters.

Autonomous loop orchestration coordinates session-status, pr-review, qa-commit, and debug skills together. It applies changes, runs validations, and commits only when all integrated checks pass or a human triage decision is required.

Error-triggered recovery is handled through continuous iteration that applies changes, runs QA validation, and escalates failures via Jidoka tiers. The loop repeats execution automatically until checks pass or the configured iteration limit is reached.

Avoid autonomous loops when automated changes risk unsafe modifications without human oversight, or when tasks lack clear success criteria. Jidoka escalation handles edge cases, but complex triage decisions may require manual intervention beyond configured limits.