Loops for Claude Code

Has a second, independent model review every change before merge so two different reviewers must agree - code only lands when both clear the bar.

Passes code between a builder and an independent reviewer subagent until the review has no blocking findings and tests pass.

Builds toward a machine-checkable contract with explicit anti-spin guardrails - stops on no progress, repeated approaches, flip-flopping, or a spend budget.

Keeps documentation aligned with the current codebase and opens a reviewable pull request on every pass.

Writes a contract of what "done" means and what evidence proves each requirement before any work starts, then refuses to claim success without that evidence.

Every few minutes while you work, makes one small verified repository improvement - a flaky test, a stale comment, a missing type - one change, one commit.

Runs the task, then pauses and sends you approve / revise / skip before anything ships - human review as a first-class queue where the stop condition is your approval.

Recovers valuable unmerged work and prunes stale branches and PRs until the repository is tidy.

Runs realistic scenarios and fixes failures until the suite passes N times in a row with no regressions.

A nightly routine that watches your open PRs, auto-fixes build failures, answers review comments in a fresh worktree, and rebases stale branches while you sleep.

Builds the next item on the plan, then runs tests, typecheck and lint, feeding every failure back as the next instruction until the build is green.

Rewrites a vague request into a precise goal - exact end state, how to verify it, what not to touch, and the stop condition - then executes against it.

Profiles and optimizes the app pass after pass until it consistently meets a defined performance budget.

Triages production errors, finds the root cause, ships a verified fix, and repeats until the error budget is clean.

Adds meaningful tests pass after pass until the suite reaches a target coverage threshold and stays green.

Converts open bug reports and issues into verified, reviewer-ready pull requests, one ticket per pass.

Adversarially challenges a design or decision until every serious objection is resolved or explicitly accepted.

Updates the changelog every night with user-relevant changes and notifies the team when it ships.