applying-tcr-with-agentslisted

# Applying TCR with Agents TCR automatically commits when tests pass and reverts when they fail. Every surviving change keeps the system green. Broken code never persists. ## Setup 1. Copy `scripts/tcr.sh` from this skill into the project root. 2. Make it executable: `chmod +x tcr.sh`. 3. Add `.agent-failures.log` to `.gitignore` (prevents failure logs from being committed by `git add -A`). 4. Verify the project has at least one git commit and a working test command. ## Incompatibility with classical TDD red-phase discipline In classical TDD, a lone failing test is written first and confirmed to fail before any implementation is written. **TCR makes that impossible:** a failing test triggers a full revert, destroying the test itself. There is no stable red phase. When using TCR, replace the red/green/refactor cycle with the rule in step 3 below: test and implementation are written together as one coherent atomic change. This is TCR's substitute for the red phase — the discipline comes from keeping the unit as small as possible, not from a separate failing-test step. Do not apply a red-first TDD workflow alongside mechanical TCR automation. ## Workflow Work in small, atomic steps. Each step includes both a test and the implementation to pass it, committed as a single unit. 1. **Reset the failure log:** `> .agent-failures.log` 2. **Identify the smallest next increment.** Pick one behavior that can be tested and implemented together. 3. **Write the test and implem