16｜Claude Code with Crazyrouter Series 16: Chapter 13: Reusing Documentation to Solve Similar Problems#

This is Part 16 of the Crazyrouter Claude Code series. This article focuses on “Claude Code with Crazyrouter Series 16: Chapter 13: Reusing Documentation to Solve Similar Problems,” with key coverage of Chapter 13: Reusing Documentation to Solve Similar Problems, 13.1 Why You Should Capture Knowledge as Documentation, and How AI Can Learn Directly from Documentation.

Unified integration convention: Claude Code / Anthropic native clients use ANTHROPIC_BASE_URL=https://cn.crazyrouter.com; OpenAI-compatible SDKs, HTTP requests, and frontend/backend applications use base_url=https://cn.crazyrouter.com/v1.

What This Article Covers#

• Who this is for: developers who are using Claude Code, preparing to connect to domestic models, or want to unify team calls through Crazyrouter.
• What you will learn: how to configure environment variables according to the Crazyrouter documentation, organize workflows, and avoid /v1/v1/... issues caused by an incorrect Base URL.
• Recommended preparation: first create a separate API Token in the Crazyrouter console, then follow the Claude Code integration documentation to complete the basic setup.

Chapter 13: Reusing Documentation to Solve Similar Problems#

13.1 Why You Should Capture Knowledge as Documentation#

Capturing knowledge as documentation allows AI to learn directly from documents, improving efficiency when solving similar problems and ensuring consistency across solutions.

AI Can Learn Directly from Documentation#

Documentation Example#

Knowledge point: how to generate product copy

Documented version:

Improving AI Efficiency When Solving Similar Problems#

Ways to Improve Efficiency#

1. Fast response: AI can quickly retrieve documentation and respond quickly
2. Accurate answers: AI can accurately understand documentation and provide accurate answers
3. Consistent output: AI can produce consistent results based on documentation
4. Continuous optimization: AI can keep learning and continuously optimize

Efficiency Improvement Example#

Scenario: generating copy for multiple products

Without documentation:

• You need to explain the requirements in detail every time
• You need to adjust the prompt every time
• You need to review the results every time
• Time required: 10–15 minutes per piece of copy

With documentation:

• AI can quickly retrieve the documentation
• AI can accurately understand the requirements
• AI can generate consistent results
• Time required: 3–5 minutes per piece of copy

Efficiency gains:

• Time saved: 60–70%
• Better quality: more consistent
• Lower cost: less manual intervention

Ensuring Consistency Across Solutions#

Why Consistency Matters#

1. Quality assurance: consistent solutions help ensure quality
2. Brand image: consistent output helps maintain brand image
3. User experience: consistent experiences improve user satisfaction
4. Easier management: consistent output is easier to manage

How to Ensure Consistency#

1. Standardized processes: establish standardized processes
2. Templated output: use templated output
3. Documented knowledge: turn knowledge into documentation
4. Regular reviews: regularly review output quality

Consistency Example#

Scenario: a team needs to generate a large volume of product copy

Without documentation:

• Everyone generates copy in a different style
• Quality is unstable each time
• A lot of manual adjustment is required
• Brand image is inconsistent

With documentation:

• Everyone generates copy based on the documentation
• The generated style is consistent
• Quality is stable
• Brand image is consistent

Easier Team Collaboration and Knowledge Sharing#

Benefits of Collaboration#

1. Knowledge sharing: team members can share knowledge
2. Experience transfer: experience can be passed on to new members
3. Less duplication: avoid duplicate work and repeated learning
4. Higher efficiency: improve overall team efficiency

Collaboration Methods#

1. Build a knowledge base: create a team knowledge base
2. Share documents: share useful documents
3. Communicate regularly: regularly exchange experience and insights
4. Keep it updated: continuously update the knowledge base

Collaboration Example#

Scenario: a team needs to handle a large volume of customer inquiries

Without documentation:

• Everyone handles cases independently
• Experience is not shared
• There is a lot of duplicate work
• Efficiency is low

With documentation:

• Build a customer inquiry knowledge base
• The team shares documents
• Experience is transferred quickly
• Efficiency improves significantly

Example Cases#

Case: Building a Product Copy Knowledge Base#

Requirement: the team needs to generate a large volume of product copy

Knowledge base setup:

1. Document structure

1. Document content

• Description of copy types
• Checklist of copy elements
• Standards for copy requirements
• Library of example cases

1. Usage workflow
2. Retrieve from the knowledge base
3. Select the appropriate template
4. Adjust based on product information
5. Generate the copy
6. Human review

Results:

• Previous time required: 10–15 minutes per piece of copy
• Current time required: 3–5 minutes per piece of copy
• Time saved: 60–70%
• Consistency improvement: significant

Tips#

1. Document promptly: document new experience as soon as you have it
2. Keep the structure clear: make the document structure clear and easy to search
3. Provide detailed content: make the documentation detailed and easy to understand
4. Include plenty of examples: provide rich examples for reference
5. Keep it updated: continuously update the documentation so it stays current

Now, start turning your knowledge into documentation!

13.2 How to Capture Knowledge as Reusable Documentation#

Learning how to capture knowledge as reusable documentation helps preserve and transfer experience while improving work efficiency.

Record the Key Steps of Successful Solutions#

Recording Principles#

1. Timeliness: record it immediately after success
2. Accuracy: accurately record every step
3. Completeness: record the complete solution
4. Reusability: make the record easy to reuse

What to Record#

1. Problem description: describe the problem clearly
2. Solution: record the solution in detail
3. Key steps: highlight the key steps
4. Notes: record important considerations
5. Result validation: record the validation results

Recording Example#

Problem: how to use Claude Code to generate high-quality product copy

Record:

Please help me generate product copy:

Product information:

• Product name: [Product name]
• Key features: [Feature 1], [Feature 2], [Feature 3]
• Target users: [Target users]

Copy requirements:

1. Highlight the product’s core selling points
2. Use vivid and engaging language that attracts the target users
3. Keep the length around [word count]
4. Style requirements: [style]
5. Include a call to action

Organize Problem Types and Corresponding Solutions#

Classification Methods#

1. By type: classify by problem type
2. By domain: classify by application domain
3. By difficulty: classify by solution difficulty
4. By frequency: classify by occurrence frequency

Classification Example#

Category 1: By type

Category 2: By domain

Write clear operating guides#

Elements of an operating guide#

1. Clear objective: Clearly define the goal of the operating guide
2. Clear steps: Make the steps easy to understand
3. Rich examples: Provide plenty of examples
4. Notes: Call out important considerations

Operating guide example#

Topic: How to use Claude Code to generate meeting minutes

Operating guide:

Please help me generate meeting minutes based on the meeting notes:

Meeting notes:

[Paste meeting notes]

Meeting minutes requirements:

1. Extract the key points of the meeting
2. Organize the decisions made
3. List action items, including owners and due dates
4. Include the next meeting schedule
5. Use standard formatting with a clear structure

Participants: Zhang San, Li Si, Wang Wu

Meeting content:

1. Discuss project progress

   • Zhang San: Frontend development is 80% complete
   • Li Si: Backend development is 70% complete
   • Wang Wu: Testing has not started yet

2. Discuss issues

   • Testing progress is behind schedule
   • Test preparation needs to be accelerated

3. Decisions

   • Hold another meeting next week
   • Wang Wu will prepare the test plan

Project Progress Meeting Minutes#

Meeting time: January 15, 2024, 14:00-16:00

Participants: Zhang San, Li Si, Wang Wu

Key meeting points#

Project progress updates#

• Frontend development: 80% complete (owner: Zhang San)
• Backend development: 70% complete (owner: Li Si)
• Testing: Not started yet (owner: Wang Wu)

Issue discussion#

• Testing progress is behind schedule
• Test preparation work needs to be accelerated

Decisions#

1. Hold a project progress follow-up meeting next week
2. Wang Wu is responsible for preparing the test plan

Action items#

• Zhang San: Complete the remaining 20% of frontend development
• Li Si: Complete the remaining 30% of backend development
• Wang Wu: Prepare the test plan and test cases
• Everyone: Hold a progress follow-up meeting next Wednesday

Next meeting#

Time: Next Wednesday at 2:00 PM

Agenda: Project progress follow-up

Save cases and best practices#

How to save cases#

1. Case description: Describe the case in detail
2. Issue analysis: Analyze the causes of the issue
3. Solution: Record the solution
4. Impact evaluation: Evaluate the effect of the solution

Case-saving example#

Case: Use Claude Code to optimize product copy

Case description:

Product copy needs to be generated for a smartwatch. The first version was not engaging enough and needs to be optimized.

Issue analysis:

• The first prompt was not specific enough
• The target users were not clearly defined
• The style requirements were not clearly defined

Solution:

Optimize the prompt and add specific requirements:

Impact evaluation:

• The generated copy improved significantly in quality
• It better matched the needs of the target users
• It included a clear call to action
• It saved 60% of the time

Best practices:

1. Make prompts specific and clear
2. Clearly define the target users
3. Clearly define the style requirements
4. Include a call to action
5. Iterate and optimize prompts

Tips#

1. Record promptly: Record a success immediately after it happens
2. Record in detail: Make records detailed so they are easy to reuse
3. Organize by category: Categorize records so they are easy to find
4. Update regularly: Update regularly to keep content current
5. Share and discuss: Share with others and learn from each other

Now, start turning your knowledge into reusable documents!

13.3 Let AI use the accumulated documents#

After turning knowledge into documents, you need to make sure AI can use those documents so they can deliver real value.

Upload documents to the Claude Code knowledge base#

Upload methods#

1. Upload individually: Upload documents one by one
2. Bulk upload: Upload multiple documents in bulk
3. Folder upload: Upload an entire folder
4. Regular updates: Update documents regularly

Upload steps#

Step 1: Prepare documents

• Organize the documents you want to upload
• Make sure the document formats are correct
• Check that the document content is complete

Step 2: Open Claude Code

• Launch Claude Code
• Open the file browser area

Step 3: Upload documents

• Click the upload file button
• Select the documents to upload
• Confirm that the upload succeeded

Step 4: Verify the upload

• Check whether the documents were uploaded successfully
• Verify that the document content is complete
• Test whether the documents are accessible

Upload example#

Document list:

1. Product Copy Generation Guide.md
2. Meeting Minutes Generation Guide.md
3. Data Analysis Guide.md
4. Weekly Report Generation Guide.md

Upload steps:

1. Open Claude Code
2. Click the upload file button
3. Select all documents
4. Confirm that the upload succeeded

Configure AI access permissions for documents#

Principles for permission configuration#

1. Security: Ensure document security
2. Accessibility: Ensure AI can access the documents
3. Flexibility: Support flexible configuration
4. Manageability: Make permissions easy to manage

Permission configuration methods#

Method 1: Global configuration

• Configure access permissions for all documents
• Suitable for public documents

Method 2: Category-based configuration

• Configure access permissions by category
• Suitable for different types of documents

Method 3: Individual configuration

• Configure access permissions for each document individually
• Suitable for sensitive documents

Permission configuration examples#

Configuration 1: Global configuration

Configuration 2: Category-based configuration

Test how well AI learns from documents#

Testing methods#

1. Functional testing: Test whether AI can access the documents
2. Accuracy testing: Test whether AI can accurately understand the documents
3. Application testing: Test whether AI can apply knowledge from the documents
4. Efficiency testing: Test AI response speed

Test example#

Test 1: Functional test question: Please help me generate product copy

Expected result: AI can retrieve the Product Copy Generation Guide and generate copy based on the guide Result: ✓ Success

Test 2: Accuracy test question: Generate copy for a smartwatch based on the product copy generation guide

Expected: The AI can accurately understand the guide requirements and generate copy that meets them

Result: ✓ Success

Test 3: Application test

Question: Please help me generate meeting minutes

Expected: The AI can apply the meeting minutes generation guide and produce high-quality meeting minutes

Result: ✓ Success

Test 4: Efficiency test question: Please help me generate a weekly report

Expected: The AI can quickly retrieve the weekly report generation guide and quickly generate the weekly report

Result: ✓ Success, response time 5 seconds

Optimize Document Structure to Improve AI Retrieval Efficiency#

Optimization Principles#

1. Clear structure: The document structure should be clear
2. Well-defined hierarchy: The document hierarchy should be easy to follow
3. Explicit keywords: Keywords should be clear
4. Complete indexes: Indexes should be comprehensive

Optimization Methods#

Method 1: Optimize the document structure

• Use clear headings
• Use a reasonable hierarchy
• Use explicit categories

Method 2: Add keywords

• Add keywords at the beginning of the document
• Add keywords in important sections
• Add keywords in examples

Method 3: Build indexes

• Build a document index
• Build a keyword index
• Build an example index

Optimization Example#

Before optimization:

After optimization:

Example Cases#

Case: Build and Use a Knowledge Base#

Requirement: Build a product copy knowledge base so the AI can generate copy based on it

Implementation steps:

Step 1: Organize knowledge

• Collect experience in generating product copy
• Organize copy types and requirements
• Summarize best practices

Step 2: Write documents

• Write a product copy generation guide
• Include detailed operating steps
• Include rich example cases

Step 3: Upload documents

• Upload the documents to Claude Code
• Verify that the upload succeeded
• Configure access permissions

Step 4: Test the results

• Test whether the AI can retrieve the documents
• Test whether the AI can apply the documents
• Test the AI's response speed

Step 5: Optimize the documents

• Optimize the documents based on the test results
• Optimize the document structure
• Add keywords and indexes

Results:

• The AI can accurately retrieve documents
• The AI can generate copy based on the documents
• Copy quality improves significantly
• Generation speed increases by 50%

Tips#

1. Clear structure: Keep the document structure clear so the AI can retrieve it easily
2. Explicit keywords: Make keywords explicit so the AI can understand them easily
3. Rich examples: Provide plenty of examples so the AI can refer to them
4. Regular updates: Update documents regularly to keep them current
5. Test and validate: Regularly test the AI's retrieval and application performance

Now, let the AI use the documents you've accumulated!

13.4 Reuse Documents to Solve Similar Problems#

The ultimate goal of documenting knowledge is reuse, so the AI can quickly solve similar problems based on the documents.

Describe New Problems to the AI#

Description Principles#

1. Clear and explicit: The problem description should be clear and explicit
2. Complete and accurate: The problem description should be complete and accurate
3. Sufficient context: Provide enough context
4. Clear goal: Clearly state the goal of solving the problem

Description Methods#

Method 1: Direct description

• Describe the problem directly
• Provide relevant background
• Explain the expected result

Method 2: Reference documents

• Reference relevant documents
• Specify the relevant sections in the documents
• Explain how to apply the documents

Method 3: Compare with references

• Compare with similar problems
• Reference similar solutions
• Explain the differences

Description Examples#

Example 1: Direct description

Example 2: Reference documents

Example 3: Compare with a reference

AI Automatically Retrieves Relevant Documents#

Retrieval Mechanism#

1. Keyword matching: Match documents based on keywords
2. Semantic understanding: Understand documents based on semantics
3. Contextual association: Associate documents based on context
4. Learning-based optimization: Optimize retrieval based on learning

Retrieval Example#

Question: Please help me generate product copy

AI retrieval process:

1. Analyze the question and extract keywords: product copy, generation
2. Retrieve relevant documents from the knowledge base
3. Find the "Product Copy Generation Guide" document
4. Extract relevant content from the document
5. Generate a solution based on the document

Retrieval result:

AI Generates Solutions by Combining Documented Experience#

Solution Generation Process#

1. Understand the problem: The AI understands the user's problem
2. Retrieve documents: The AI retrieves relevant documents
3. Apply knowledge: The AI applies knowledge from the documents
4. Generate a solution: The AI generates a solution
5. Optimize and adjust: The AI optimizes and adjusts based on feedback

Solution Generation Example#

Question: Please help me generate product copy for a smart speaker

AI generation process:

1. Understand the problem: Product copy needs to be generated for a smart speaker
2. Retrieve documents: Retrieve the "Product Copy Generation Guide"
3. Apply knowledge: Apply the copy elements and requirements from the document
4. Generate a solution: Generate product copy
5. Optimize and adjust: Optimize based on user feedback

Generated solution:

Validate the Solution Results#

Validation Methods#

1. Accuracy validation: Validate whether the solution is accurate
2. Completeness validation: Validate whether the solution is complete
3. Practicality validation: Validate whether the solution is practical
4. Efficiency validation: Validate whether the solution is efficient

Validation Examples#

Validation 1: Accuracy validation

• Check whether the copy accurately reflects the product features
• Check whether the copy meets the needs of the target users
• Check whether the copy includes a call to action
• Result: ✓ Accurate

Validation 2: Completeness validation

• Check whether the copy includes all product features
• Check whether the copy meets the word count requirement
• Check whether the copy meets the style requirement
• Result: ✓ Complete

Validation 3: Practicality validation

• Check whether the copy can be used directly
• Check whether the copy requires major revisions
• Check whether the copy achieves the expected results
• Result: ✓ Practical

Validation 4: Efficiency validation

• Check whether the generation time is reasonable
• Check whether time was saved
• Check whether efficiency was improved
• Result: ✓ Efficient, saves 70% of the time

Example Cases#

Case: Reuse Documents to Solve Multiple Similar Problems#

Requirement: Generate copy for multiple products

Document: Product Copy Generation Guide.md Question 1: Generate copy for a smartwatch

Description: Please help me generate product copy for a smartwatch.

AI retrieval: Retrieved the product copy generation guide.

AI generation: Generated copy based on the documentation.

Validation: The copy is accurate, complete, practical, and efficient.

Result: ✓ Successful

Question 2: Generate copy for wireless earbuds

Description: Please help me generate product copy for wireless earbuds.

AI retrieval: Retrieved the product copy generation guide.

AI generation: Generated copy based on the documentation.

Validation: The copy is accurate, complete, practical, and efficient.

Result: ✓ Successful

Question 3: Generate copy for a portable speaker

Description: Please help me generate product copy for a portable speaker.

AI retrieval: Retrieved the product copy generation guide.

AI generation: Generated copy based on the documentation.

Validation: The copy is accurate, complete, practical, and efficient.

Result: ✓ Successful

Impact:

• Time to generate each piece of copy: 3-5 minutes
• Time saved compared with the previous process: 60-70%
• Copy quality: Significantly improved
• Consistency: Significantly improved

Tips#

1. Be clear in your description: Make the problem description clear so the AI can understand it easily.
2. Reference documentation: Reference relevant documentation to improve retrieval accuracy.
3. Provide context: Provide enough context so the AI can apply it effectively.
4. Validate results: Validate the effectiveness of the solution to ensure quality.
5. Provide feedback and optimize: Provide feedback to help the AI improve.

Now let the AI solve your problems based on your documentation!

13.5 Continuously Update and Iterate on Documentation#

Documentation is not static. It needs to be continuously updated and iterated based on new cases and new experience to retain its value.

Update Documentation Based on New Cases#

Triggers for Updates#

1. New cases appear: A new successful case appears.
2. New methods are discovered: A new solution is found.
3. New problems appear: A new type of problem appears.
4. User feedback: User feedback is received.

Update Methods#

Method 1: Add a new case

• Add the new case to the documentation.
• Describe the case background.
• Describe the case solution.
• Describe the case results.

Method 2: Update a method

• Update an existing solution.
• Optimize the method steps.
• Add more explanation for the method.

Method 3: Add a new method

• Add a new solution.
• Describe where the new method applies.
• Describe the advantages of the new method.

Update Example#

Before the update:

After the update:

Optimize Documentation Content to Improve AI Understanding Efficiency#

Optimization Principles#

1. Clear structure: The documentation structure should be clear.
2. Distinct hierarchy: The documentation hierarchy should be easy to follow.
3. Clear keywords: Keywords should be explicit.
4. Rich examples: Examples should be comprehensive.

Optimization Methods#

Method 1: Optimize the structure

• Use clear headings.
• Use a reasonable hierarchy.
• Use explicit categories.

Method 2: Add keywords

• Add keywords at the beginning of the documentation.
• Add keywords in important sections.
• Add keywords in cases.

Method 3: Enrich examples

• Add more cases.
• Add different types of cases.
• Add cases for different scenarios.

Optimization Example#

Before optimization:

After optimization:

Please help me generate product copy:

Product information:

• Product name: [Product name]
• Key features: [Feature 1], [Feature 2], [Feature 3]
• Target users: [Target users]

Copy requirements:

1. Highlight the product's core selling points.
2. Use vivid and engaging language to attract the target users.
3. Keep the length around [word count].
4. Style requirement: [Style]
5. Include a call to action.

Regularly Clean Up Outdated Content#

Cleanup Principles#

1. Timeliness: Clean up outdated content promptly.
2. Accuracy: Ensure the content is accurate.
3. Relevance: Ensure the content is relevant.
4. Practicality: Ensure the content is useful.

Cleanup Methods#

Method 1: Delete outdated content

• Delete outdated cases.
• Delete outdated methods.
• Delete outdated explanations.

Method 2: Update outdated content

• Update outdated information.
• Update outdated methods.
• Update outdated cases.

Method 3: Mark outdated content

• Mark outdated content.
• Explain why it is outdated.
• Provide alternatives.

Cleanup Example#

Before cleanup:

After cleanup:

Establish a Documentation Update Mechanism#

Designing the Update Mechanism#

1. Regular updates: Check and update documentation regularly.
2. Triggered updates: Trigger updates based on specific events.
3. Version management: Manage documentation versions.
4. Update records: Record update history.

Examples of Update Mechanisms#

Mechanism 1: Regular updates

• Check documentation once a month.
• Check whether there are new cases.
• Check whether there are new methods.
• Check whether there is user feedback.

Mechanism 2: Triggered updates

• Update when user feedback is received.
• Update when a new method is discovered.
• Update when a new case appears.
• Update when an error is found.

Mechanism 3: Version management

• Use version numbers to manage documentation.
• Record the content of each update.
• Keep historical versions.
• Support version rollback.

Mechanism 4: Update records

• Record the update time.
• Record the update content.
• Record the reason for the update.
• Record the impact of the update.

Example Case#

Case: Continuously Optimizing the Product Copy Generation Guide#

Initial version:

• Basic steps
• 1 example case
• Simple explanation

First update:

• Added 2 new cases
• Optimized the steps
• Added best practices

Second update:

• Added keywords
• Optimized the documentation structure
• Added more cases

Third update:

• Deleted outdated cases
• Updated outdated methods
• Added new methods

Final version:

• Clear documentation structure
• Rich example cases
• Detailed steps
• Complete best practices

Impact:

• AI retrieval efficiency improved by 50%
• Copy generation quality improved by 30%
• User satisfaction improved by 40%

Tips#

1. Update promptly: Update documentation promptly when new content is available.
2. Check regularly: Regularly check documentation accuracy.
3. Version management: Manage documentation versions.
4. Record updates: Record update history.
5. User feedback: Collect user feedback and keep optimizing. Now start continuously updating and iterating on your documentation!

---

• Previous: Article 15
• Next: Article 17
• Configuration docs: Connect Claude Code to Crazyrouter
• API endpoint guide: Base URL and /v1 Usage
• One-click setup script: Crazyrouter Claude Code One-Click Setup Script

Start Connecting to Crazyrouter#

If you're ready to connect Claude Code, domestic models, or your own applications to Crazyrouter through a unified entry point, follow this sequence:

1. Go to the Crazyrouter console, create a dedicated API Token, and manage permissions separately by project or team.
2. Claude Code uses the root domain: https://cn.crazyrouter.com; OpenAI-compatible SDKs use: https://cn.crazyrouter.com/v1.
3. When you need to automatically check the environment or quickly write configuration, use the Crazyrouter Claude Code One-Click Setup Script.
4. If debugging fails, check the console logs first, then verify the API Endpoint guide. Pay special attention to whether the Base URL has an extra /v1.

When you need to evaluate model costs or choose different models, start with the Crazyrouter pricing and models page, then add your commonly used models to the Token allowlist.