/wbCheck — Examples

/wbCheck is the only command you run BEFORE another command. It's a gate, not a task. ⚠️ Examples use a fictional package "wb-demo" to avoid leaking real answer-key data.

---

/wbCheck vs other commands

/wbCheck	/wbTrack	/wbReview	/wbAudit
"Does the model understand my code?"	"Log every command with commentary"	"Did the model do what the plan said?"	"Is this code ready to ship?"
Runs BEFORE the task	Runs DURING the session	Runs AFTER the task	Runs independently
Tests the MODEL	Tracks the SESSION	Tests the OUTPUT	Tests the CODE
30-second quiz	Toggle on/off	5-minute inspection	15-minute deep dive

---

Example 1 — wbcheck (no args) → Monorepo quiz

Terminal:

$ wbcheck

Generates 6 monorepo-wide questions. You copy them.

Model chat (OpenCode with Qwen3):

You paste the /wbCheck prompt with the 6 questions. Qwen reads the workspace, answers, and ends with:

"I have answered all the quiz questions. Did I pass? (yes/no)"

Score: 5/6 = ✅ PASS

You: yes
Model: ✅ Verified and ready. Send me any /wb* command.

That's it. No walkthrough question. No extra steps. The model is verified and you start working.

---

Example 2 — wbcheck wb-demo wbTest → Targeted quiz → Quick start

Terminal:

$ wbcheck wb-demo wbTest

6 questions (3 identity + 3 testing).

Model chat: Model answers → asks "Did I pass?" → 6/6 → You say "yes" → ✅ Verified.

Next: You send /wbTest packages/wb-demo directly.

---

Example 3 — wbcheck wb-demo wbAudit → FAIL

Terminal:

$ wbcheck wb-demo wbAudit

Model chat (GPT 5 Nano):

• Hallucinated the framework version → ❌
• Fabricated a sanitization approach → ❌

Model asks: "Did I pass?" → You say: no

Model: "Understood. This session ends here."

You close the window. Switch to a stronger model.

---

Example 4 — /wbCheck + /wbTrack (independent combo)

Scenario: New model, long documented session.

Step 1: wbcheck wb-core ← terminal: generate questions
Step 2: Paste to model ← model answers quiz
Step 3: You grade → PASS → "yes" ← model is verified
Step 4: /wbTrack packages/wb-core ← start tracking (separate command)
Step 5: /wbTest wb-core ← creates report + §1 in session
Step 6: /wbStopTrack ← ends session, extracts derivatives

Key: /wbCheck and /wbTrack are independent commands. You can use one without the other.

---

The Pattern

Every /wbCheck answers three meta-questions:

1. Does the model know WHAT this is? (identity questions)
2. Does the model know HOW it works? (architecture/testing/build)
3. Can it give SPECIFIC facts, not generic statements? (numbers, filenames, exact versions)

If the answers are specific and correct → the model read your code. If the answers are vague or wrong → the model is guessing from training data.

---

Basic Usage

# Standard command execution
/wbCheck frontEnd/wbc-ui3/packages/

# Execution with explicit target ID filter
/wbCheck deployement/apps/wb-jobs/ --id=1,2,3

Model Fallback Chain (.wb/bin/wbRun)

When executing CLI dispatches, /wbCheck wraps binary calls in .wb/bin/wbRun to ensure output-error guarding:

.wb/bin/wbRun claude -p --permission-mode auto "/wbCheck target/" || .wb/bin/wbRun agy --model gemini-3.1-pro-high -p "/wbCheck target/" || .wb/bin/wbRun opencode run -m opencode-go/deepseek-v4-pro "/wbCheck target/"

---

Role Model Overrides (--planner, --worker, --validator, --mechanical)

You can override default models per role directly from the command line:

# Override Worker and Validator models
/wbCheck packages/ --wave=A --worker="DeepSeek V4 Pro,Kimi K3" --validator="Gemini 3.5 Pro"

# Override all 4 roles simultaneously
/wbCheck apps/wb-jobs/ --wave=all --planner="Claude Opus 5" --worker="DeepSeek V4 Pro" --validator="Claude Sonnet 4.7" --mechanical="Gemini 3 Flash"

Autonomous Non-Interactive Execution (-y / --yes)

Run /wbCheck in zero-touch print mode without stopping for manual decision prompts:

.wb/bin/wbRun claude -p --permission-mode auto "/wbCheck frontEnd/wbc-ui3/packages/ --wave --as='expert,steps' -y"