Business Review

Objective

Core Principles

• Read and write the

  opc-doc/

  directory under the current working directory by default
• In teaching mode, first explain that a review is not a running account, but a process to identify bottlenecks
• Ask only one question at a time by default; if several questions are very lightweight and closely related to each other, they can be combined into 2 to 3 questions
• Provide 3 bottleneck hypotheses or priority focuses by default, with an additional option

  4. I have my own solution

• Write formal results only after user confirmation
• Do not give direct recommendation conclusions, only conduct solution analysis

Mandatory Completion Items for This Step

1. Current operating status summary
2. Major bottleneck identification
3. Confirmation of priority focuses for the next cycle

Priority Confirmation Order

1. The most stuck part recently
2. The most possible bottleneck type
3. Which point to prioritize for correction in the next cycle

Completion Criteria

• 3 bottleneck hypotheses or priority focuses have been formed
• The user has confirmed the most realistic current bottleneck

Content to Explain in This Step

• A review focuses on demand, conversion, assets and risks
• The goal is not to prove you have worked hard, but to find the most worthwhile point to correct
• This step will determine the sole focus of the next cycle

Trigger Conditions

• You feel stuck after operating for a period of time and don't know where the problem lies
• You want to conduct a structured review instead of continuing to operate based on intuition
• There is a significant decline in conversion, retention or transaction volume
• The user voluntarily chooses to conduct a review

Input

• opc-doc/outputs/**

  (all foundation building period products, for comparison)

• opc-doc/reviews/dashboard.json

  (historical review records, if available)

• opc-doc/outputs/09-dashboard-review/

  (previous reviews in this cycle, if available)

opc-doc/

• If the user has no operation experience at all, do not conduct a fake review, remind the user to execute first before doing a review

Execution Steps

1. Explain the goal of this step
2. Ask only one question at a time by default; if several questions are very lightweight and closely related to each other, they can be combined into 2 to 3 questions, for example:
   • What is the most stuck step for you recently?
   • Are you facing no traffic, no purchases, or no reuse of completed work?
3. Give brief feedback after each round of answers
4. Generate 3 bottleneck judgments, for example:
   • Insufficient demand verification
   • Unsmooth conversion path
   • Too weak asset accumulation
5. Explain the applicable scenarios, basis and cost of each judgment
6. Add the option

   4. I have my own solution

   by default
7. Let the user confirm, combine, modify, or directly propose their own version
8. Write the formal review result only after user confirmation

Output

1. Explanation of this step
2. Current operating status summary
3. 3 bottleneck judgments +

   4. I have my own solution

4. Applicable scenarios, basis and cost of each judgment
5. Request for user confirmation or modification

Persistence Checkpoint (must be completed after each trigger)

• opc-doc/outputs/09-dashboard-review/review-[YYYYMMDD].md

  (operating status summary + bottleneck judgment + sole priority focus for the next cycle)

• opc-doc/reviews/dashboard.json

  (structured review data, append write, retain history, no overwriting)

• opc-doc/state/current-stage.json

  (write:

  {"stage": "09-dashboard-review", "status": "completed", "next_stage": "execution-or-回退阶段", "summary": "一句话核心瓶颈和下一重点"}

  )

• opc-doc/state/decisions.json

  (append the sole priority focus of the next cycle, indicate the date)

"✅ This review has been saved. According to the review conclusion: [briefly explain whether the next step is to continue execution, roll back to a certain stage, or trigger asset accumulation]"

When to Call Other Skills

• If the review finds gaps in the pre-stage, clearly recommend rolling back to the corresponding skill
• Do not ask the user to repeat the entire process without basis

Exception Handling

• If the data is incomplete, only conduct a lightweight qualitative review, do not pretend to have a complete conclusion
• Do not give multiple next-step focuses at the same time, maintain a single priority