auto-issues

You automate an issue workflow by running sequential phase orchestrations for research, plan, implement, and update inside an isolated worktree. The pipeline enforces strict artifact validation and completes with a single-session inline git add, commit, and push.

An issue pipeline orchestration sequentially executes research, planning, implementation, and updating phases to produce required engineering artifacts. It validates these artifacts against minimum content thresholds and writes per-phase logs to ensure a strict audit trail for the code review process.

Yes, issue pipeline automation requires a dedicated git branch named worktree-issue-<N>. The pipeline derives the issue number directly from this branch name to keep execution isolated and correctly map the automated phases to the specific issue.

Yes, you can use Claude and bash scripts to orchestrate multi-phase engineering tasks through per-phase Claude Codex invocations. The pipeline uses bash scripts to manage the sequential execution, generate an eval verdict, and validate logs and artifacts produced during each phase.

You validate artifact integrity and logs during automated issue resolution by enforcing artifact existence and minimum content thresholds at each phase. The pipeline writes per-phase logs to a tasks/logs directory and generates a final evaluation verdict using a dedicated validation script.

Automating issue pipelines inside isolated worktrees requires strict sequential phase orchestration and an existing worktree-issue-<N> branch. The pipeline cannot run without deriving the issue number from the branch name and depends on per-phase Claude Codex invocations for execution.