Align verification docs with CLI surface

Author: chenbo

docs/concepts/architecture.md

@@ -450,10 +450,10 @@ The report still includes retrieval signals such as Recall@K, Precision@K, token
 npm run build
 npm run check
 npm run lint
-npm run format
 npm run format:check
-npm test
+npm run docs:cli:check
+npm run test
 npm run benchmark
-npm run build
+npm run benchmark:agent
 npm run pack:dry-run
 ```

docs/concepts/loop-engineering.md

@@ -194,9 +194,12 @@ Each loop writes a durable iteration directory:
 The runtime is exposed through CLI commands, including:
 
 ```txt
-build, run, loop, plan, pack, verify, task, tests, impact,
-policy, validate-contracts, freshness, drift, delta, evolve,
-trace start/add/run/show, rag search, benchmark, retrieve, diff, update, explain
+sidecar, report, status, doctor, opencode/oc run, build, run,
+loop, plan, pack, verify, task, tests, impact, policy,
+hallucination, regression, validate-contracts, freshness, drift,
+delta, evolve, trace start/add/run/show, rag search, retrieve,
+orchestrate, agent run, benchmark, benchmark-agent, diff, update,
+explain
 ```
 
 The follow-up release check is to keep the npm package, `dist/`, CLI help, and docs aligned with this surface.

docs/concepts/loop-engineering.zh-CN.md

@@ -370,6 +370,11 @@ build/refresh pack
 当前 `src/cli/index.ts` 已经注册了主要 Harness 命令，包括：
 
 ```txt
+sidecar
+report
+status
+doctor
+opencode / oc run
 build
 savings
 rag export
@@ -380,6 +385,8 @@ graph
 readiness
 validate
 policy
+hallucination
+regression
 validate-contracts
 freshness
 drift
@@ -393,7 +400,10 @@ verify
 task
 tests
 impact
+orchestrate
+agent run
 benchmark
+benchmark-agent
 retrieve
 diff
 update