helix-implement-reviewlisted

# Helix implement-review loop Use this loop when the implementation direction is already settled and a separate planning phase would add no value. ## Required launch protocol 1. Treat `$ARGUMENTS` as the exact task. If it is empty, ask for the task and stop. 2. Read repository rules, inspect the target, and identify the expected writer scope plus canonical verification argv vector. Do not encode shell syntax. The first item must be a PATH-resolved executable name without `/`, `\\`, or a leading dash; pass repository scripts to `node` or another interpreter instead of executing `./…` directly. 3. Select `mode`: `original` by default, or explicit `graph`. Original uses `${CLAUDE_PLUGIN_ROOT}/workflows/helix-implement-review.js`; graph uses `${CLAUDE_PLUGIN_ROOT}/workflows/graph/helix-implement-review.js`. If a supplied mode is neither exact `original` nor exact `graph`, stop before doctor, evidence-session, or Workflow invocation. Configure `maxPasses` (`1..5`, default `3`) and optional role models for `builder`, `tester`, `documenter`, `reviewer`, `redteam`, and `verifier`. Omitted roles inherit the session model. 4. Run `node "${CLAUDE_PLUGIN_ROOT}/bin/helix-cc-doctor" --json`. Continue only when `paths.nativeClaudeWorkflow.locallyReady` or `paths.cliProxyNativeWorkflow.locallyReady` is true. Stop when a global subagent-model override is present and per-role models were requested. 5. In a CLIProxyAPI session, require every explicit model to appear exac