Feature Request: Auto-detect and use project knowledge bases

[markusbucher]

Select Topic Area

Product Feedback

Copilot Feature Area

VS Code

Body

Problem Statement

AI coding assistants (GitHub Copilot, Claude, etc.) do not automatically detect and utilize project-specific knowledge bases, even when these knowledge bases are explicitly designed for AI consumption and follow clear conventions.

Real-World Impact

Our team has invested significant effort in building a comprehensive AI knowledge management system:

• 51+ repository analysis documents
• Structured architecture documentation
• Domain knowledge and patterns
• API documentation and deployment guides
• Scripts to load context for AI assistants

Despite this investment, the AI assistant ignored all of this context and generated generic responses that didn't align with our actual codebase.

---

Current Behavior

When working in a workspace, AI assistants:

1. ❌ Do not automatically check for knowledge base directories
2. ❌ Do not read context files like COPILOT_CONTEXT.md
3. ❌ Do not recognize copilot_knowledge/ conventions
4. ❌ Do not prompt users about available knowledge bases
5. ❌ Generate responses without project-specific context

Example Scenario

Workspace Structure:

project/
├── COPILOT_CONTEXT.md              # Context file for AI
├── copilot_knowledge/              # Structured knowledge base
│   ├── architecture/
│   ├── repositories/
│   ├── apis/
│   └── README.md
├── docs/COPILOT_KNOWLEDGE_GUIDE.md # Instructions for using knowledge base
└── scripts/copilot_context.sh     # Script to load context

User Request: "Create POC requirements for AI agent working on PRODUCTNAME"

AI Behavior:

• Generates generic requirements
• Uses placeholder names and fake examples
• Ignores 50+ pages of actual system documentation
• User must explicitly say "check the knowledge base"

---

Expected Behavior

AI assistants should:

1. Auto-Detection on Workspace Open

✅ Workspace opened: /Users/dev/eh2.0
✅ Detected: COPILOT_CONTEXT.md
✅ Detected: copilot_knowledge/ (52 files)
✅ Detected: docs/COPILOT_KNOWLEDGE_GUIDE.md

💡 Notification to user:
   "I found a knowledge base with 52 documentation files about 
    your PRODUCT system. Should I use this context for all 
    responses? [Yes] [Not now] [Never]"

2. Automatic Context Loading

When standard context files exist, automatically read:

• COPILOT_CONTEXT.md (current working context)
• copilot_knowledge/README.md (knowledge base overview)
• .github/copilot-instructions.md (if exists)
• Relevant files from knowledge base based on task

3. Context-Aware Responses

Before generating code/documentation:

1. Check if relevant knowledge exists
2. Search knowledge base for related concepts
3. Use actual system names, patterns, and architectures
4. Cite knowledge base sources in responses

4. Proactive Knowledge Base Querying

When user asks about the system:

User: "How does event routing work?"

AI Process:
1. Search copilot_knowledge/architecture/ for "event routing"
2. Find: beb_domain_knowledge.md, system_overview.md
3. Read relevant sections
4. Generate response based on actual architecture
5. Cite sources: "[See: architecture/beb_domain_knowledge.md]"

---

Proposed Solution

For GitHub Copilot

A. Standard Knowledge Base Conventions

Recognize these patterns automatically:

• copilot_knowledge/ or .copilot/
• COPILOT_CONTEXT.md
• .github/copilot-instructions.md
• docs/COPILOT_*.md

B. VS Code Extension Enhancement

Add Copilot settings:

{
  "github.copilot.knowledgeBase.enabled": true,
  "github.copilot.knowledgeBase.autoDetect": true,
  "github.copilot.knowledgeBase.paths": [
    "copilot_knowledge/",
    "COPILOT_CONTEXT.md"
  ]
}

C. Chat Interface Enhancement

When knowledge base detected:

• Show indicator: 🧠 "Using project knowledge base"
• Allow toggling: "Use knowledge base: [ON/OFF]"
• Show sources: "Based on: architecture/system_overview.md"

For Claude / AI Assistants

A. Workspace Analysis Behavior

On first interaction in a workspace:

1. Scan for common knowledge base patterns
2. Read index/README files
3. Offer to use knowledge base
4. Remember preference for session

B. Tool Enhancement

Add explicit tool for knowledge base awareness:

{
  "name": "check_project_knowledge_base",
  "description": "Automatically invoked on workspace open to detect knowledge bases",
  "auto_trigger": true
}

C. Best Practice Enforcement

Make it a best practice that AI assistants MUST:

• Check for context files before responding
• Ask user if context should be used
• Prefer project-specific knowledge over general knowledge

---

Benefits

For Developers

• ✅ AI responses aligned with actual codebase
• ✅ Reduced back-and-forth ("No, we use X not Y")
• ✅ Accurate code generation from day one
• ✅ Knowledge base investment pays off

For Organizations

• ✅ Consistent AI behavior across teams
• ✅ Onboarding acceleration for new developers
• ✅ Reduced hallucinations and incorrect suggestions
• ✅ Better ROI on documentation efforts

For AI Products

• ✅ Better user experience and trust
• ✅ Differentiation from competitors
• ✅ Reduced user frustration
• ✅ More accurate, valuable outputs

---

Related Standards and Conventions

Emerging Conventions in Industry

• .ai/ directory for AI-specific context
• CONTEXT.md files for Claude projects
• .github/copilot-instructions.md for Copilot
• RAG (Retrieval Augmented Generation) patterns

Similar Features in Other Tools

• Cursor IDE: Project-level context with @docs
• Codeium: Workspace-aware completions
• Tabnine: Team knowledge learning
• JetBrains AI: Project context indexing

---

Workarounds (Until Fixed)

For Users

1. Explicit Instruction in Every Chat:

   • "First, check COPILOT_CONTEXT.md and copilot_knowledge/"

2. Workspace-Level Settings:

   • Add reminder in workspace settings/README

3. Pre-prompt Engineering:

   • Start every session with context loading command

For Tool Developers

1. Build Custom Extensions:

   • VS Code extension that auto-loads context

2. RAG Integration:

   • Implement custom RAG layer on top of AI API

3. Prompt Templates:

   • Prepend knowledge base content to all prompts

---

Priority Justification

High Impact

• Affects every developer using AI assistants
• Impacts quality of AI-generated code/docs
• Reduces effectiveness of knowledge base investments

High Frequency

• Every AI interaction could benefit from context
• Multi-repository projects especially affected
• Enterprise teams with complex codebases most impacted

Implementation Feasibility

• Standard file patterns are easy to detect
• File reading capability already exists
• UI changes are minimal (indicator, toggle)
• Can be opt-in initially

---

Submission Information

For GitHub Copilot

• Submit to: https://github.com/github/feedback/discussions (Copilot category)
• Title: "Feature Request: Auto-detect and use project knowledge bases"
• Labels: enhancement, copilot-chat, ux-improvement

For Anthropic (Claude)

• Submit to: https://www.anthropic.com/feedback or support channels
• Title: "Feature Request: Workspace knowledge base auto-detection"
• Category: Product Enhancement

For VS Code

• Submit to: https://github.com/microsoft/vscode/issues
• Title: "API: Expose knowledge base conventions for AI extensions"
• Labels: feature-request, api-proposal

---

Additional Context

Our Knowledge Base System

Repository: Multi-Repository Workspace

• 51 analyzed repositories
• 450+ pages of documentation
• Automated analysis and update scripts
• Designed specifically for AI consumption

Example Knowledge Base Structure

copilot_knowledge/
├── README.md                      # Overview and navigation
├── architecture/
│   ├── system_overview.md        # High-level architecture
│   ├── beb_domain_knowledge.md   # Domain patterns
│   └── relationship_matrix.md    # Service dependencies
├── repositories/
│   ├── backend/                  # 28 Go services analyzed
│   ├── operators/                # Kubernetes operators
│   └── common/                   # Shared libraries
├── apis/
│   └── api_overview.md           # API documentation
├── deployment/
│   └── deployment_guide.md       # GitOps and deployment
└── load_context.sh               # Context loader script

---

Conclusion

AI coding assistants are incredibly powerful, but their value is dramatically reduced when they ignore project-specific knowledge that developers have carefully curated. Auto-detecting and using knowledge bases should be a default behavior, not something users need to remember to request.

This feature would:

• Improve developer productivity
• Increase trust in AI tools
• Make knowledge base investments worthwhile
• Differentiate products in competitive AI market

We strongly encourage implementing this feature and are happy to provide additional feedback, testing, or examples of our knowledge base system.

---

Contact: Markus Bucher for an SAP development org
Feedback Welcome: We're happy to discuss implementation details or provide access to our knowledge base system as a reference implementation.

[github-actions[bot]]

💬 Your Product Feedback Has Been Submitted 🎉

Thank you for taking the time to share your insights with us! Your feedback is invaluable as we build a better GitHub experience for all our users.

Here's what you can expect moving forward ⏩

• Your input will be carefully reviewed and cataloged by members of our product teams. 
  • Due to the high volume of submissions, we may not always be able to provide individual responses.
  • Rest assured, your feedback will help chart our course for product improvements.
• Other users may engage with your post, sharing their own perspectives or experiences. 
• GitHub staff may reach out for further clarification or insight. 
  • We may 'Answer' your discussion if there is a current solution, workaround, or roadmap/changelog post related to the feedback.

Where to look to see what's shipping 👀

• Read the Changelog for real-time updates on the latest GitHub features, enhancements, and calls for feedback.
• Explore our Product Roadmap, which details upcoming major releases and initiatives.

What you can do in the meantime 💻

• Upvote and comment on other user feedback Discussions that resonate with you.
• Add more information at any point! Useful details include: use cases, relevant labels, desired outcomes, and any accompanying screenshots.

As a member of the GitHub community, your participation is essential. While we can't promise that every suggestion will be implemented, we want to emphasize that your feedback is instrumental in guiding our decisions and priorities.

Thank you once again for your contribution to making GitHub even better! We're grateful for your ongoing support and collaboration in shaping the future of our platform. ⭐