Feature Request: MCP to access Claude.ai chat history and Projects

This!

I also often 'brainstorm' away from my pc. While in bath, walking, on the bus, when an idea pops in my head.
Often i use claude app or claude.ai to work these throughm but when i get home and work with claude code, i need to start copy/pasting all parts, files, snippets etc.
It would be immensely helpful if you could give claude.ai chats as an artifact to work from (by referencing their id or something)

I came to something like this as an idea (with help of claude)

Feature Request: Import Claude.ai Conversations by Chat ID

Summary

Add ability to import and reference Claude.ai conversations by their chat ID directly into Claude Code sessions, enabling seamless workflow between mobile/web brainstorming and local development.

Problem Statement

Currently, there's significant friction in the workflow between architectural planning and implementation:

1. Brainstorming phase: Users often discuss architecture, design decisions, and technical approaches on claude.ai (mobile app, web interface) while away from their development environment (commuting, relaxing, etc.)

1. Implementation phase: When returning to their desk, users must manually copy/paste all context, code snippets, architectural decisions, and conversation history from claude.ai into Claude Code

1. Context loss: Manual copying is tedious, error-prone, and often results in lost context or incomplete information transfer

Proposed Solution

Basic Implementation

Allow users to import claude.ai conversations by chat ID:

# Import entire conversation as context
claude --import-chat <chat-id>

# Or reference in a prompt
claude "implement the authentication system we discussed in chat <chat-id>"

Advanced Implementation

# Import specific messages from a conversation
claude --import-chat <chat-id> --messages 10-25

# Import and continue the conversation
claude --continue-from-chat <chat-id>

# Import as reference without making it primary context
claude --reference-chat <chat-id> "now implement the API endpoints"

Use Cases

Use Case 1: Architecture Design → Implementation

Mobile (claude.ai):
- User discusses database schema design
- Explores different approaches
- Decides on final architecture
- Creates example code snippets

Desktop (Claude Code):
$ claude --import-chat abc123xyz "implement the schema we designed"
→ Full context automatically available
→ No manual copying needed

Use Case 2: Problem Solving → Fix Implementation

Web (claude.ai):
- User debugs a complex issue
- Tries different approaches
- Finds the solution

Desktop (Claude Code):
$ claude --import-chat def456uvw "apply the fix we discovered"
→ All debugging context preserved
→ Solution implemented with full understanding

Use Case 3: Requirements Gathering → Development

Mobile (claude.ai):
- Product discussions with stakeholders
- Feature requirements refined
- Technical constraints identified

Desktop (Claude Code):
$ claude --import-chat ghi789rst "create tickets for these features"
→ Requirements automatically inform implementation
→ No context lost in translation

Benefits

1. Seamless Workflow: Natural transition from thinking/planning to implementation
2. Zero Manual Copying: Eliminates tedious copy-paste between platforms
3. Complete Context Preservation: Entire conversation history informs implementation decisions
4. Mobile-First Design Thinking: Enables architectural work anywhere, implementation at desk
5. Reduced Friction: Removes barrier between ideation and execution
6. Better Outcomes: Full context leads to implementations that better match original intent

Technical Considerations

Authentication

• Require user authentication to access their claude.ai conversations
• OAuth or API key integration with Anthropic account
• Respect existing conversation permissions and privacy

Privacy & Security

• Only allow import of user's own conversations
• Respect conversation privacy settings
• Option to import without persisting locally (ephemeral context)

Rate Limiting

• Implement reasonable rate limits for conversation imports
• Cache imported conversations locally to avoid repeated API calls

Conversation Format

• Import full message history with proper role attribution
• Preserve artifacts, code blocks, and formatting
• Support both complete import and selective message ranges

Implementation Suggestions

Phase 1: Basic Import

• Single command to import entire conversation
• Conversation becomes part of session context
• Simple chat ID lookup

Phase 2: Selective Import

• Import specific message ranges
• Reference multiple conversations
• Merge contexts intelligently

Phase 3: Smart Context

• Automatically identify relevant conversations
• Suggest importing related past conversations
• Context continuation across sessions

Alternative Solutions Considered

1. Manual copy-paste: Current workaround, but tedious and error-prone
2. Export/import via files: Requires manual download/upload, still friction
3. Share links: Doesn't provide programmatic access or context injection

None of these alternatives provide the seamless workflow that direct chat ID import would enable.

Related Features

This could potentially integrate with or enable:

• Conversation history search across platforms
• Cross-session context continuity
• Team collaboration (import shared conversations)
• Knowledge base building from past conversations

Success Metrics

If implemented successfully, we'd expect to see:

• Reduced time between design discussion and implementation
• Higher quality implementations (more complete context)
• Increased usage of claude.ai for architectural planning
• Smoother developer workflow satisfaction

Example Workflow

Before (Current):

1. Discuss architecture on claude.ai mobile (15 min)
2. Return to desk
3. Open claude.ai web, scroll through conversation (5 min)
4. Copy relevant parts (10 min)
5. Paste into Claude Code (2 min)
6. Start implementation (actual work begins)

Total overhead: ~17 minutes + context switching friction

After (With Feature):

1. Discuss architecture on claude.ai mobile (15 min)
2. Return to desk
3. claude --import-chat <id> "implement this"
4. Start implementation (actual work begins)

Total overhead: ~30 seconds

Community Impact

This feature would particularly benefit:

• Mobile-first developers: Those who do thinking work away from desk
• Consultants: Who gather requirements with clients on-the-go
• Architects: Who think through designs before implementation
• Distributed teams: Who discuss asynchronously across time zones
• Anyone with a commute: Turn transit time into productive design time

---

Priority: High - addresses significant workflow friction
Complexity: Medium - requires Anthropic API integration
Impact: High - dramatically improves cross-platform workflow

Strong +1 on this. The disconnect between claude.ai and Claude Code is a major pain point.

The problem in practice:

I'm a Max subscriber. On claude.ai, Claude knows:

• My projects and their architecture
• My preferences for code style
• My technical background
• Context from hundreds of past conversations

When I open Claude Code, it's a stranger. Same subscription, same user, completely different Claude.

This shouldn't require MCP complexity. At minimum:

1. Read-only memory sync - Claude Code pulls from claude.ai's memory at session start
2. Conversation reference - Ability to reference/continue claude.ai conversations
3. Unified memory backend - Both products use the same memory system

The alternative ask:

If integrating the memory systems is complex, at least update the ToS to allow subscribers to access their own claude.ai sessions programmatically. Let us build the bridge ourselves while you work on official integration.

We're paying for ONE Claude. We should get ONE Claude across all interfaces.

This issue has been inactive for 30 days. If the issue is still occurring, please comment to let us know. Otherwise, this issue will be automatically closed in 30 days for housekeeping purposes.

I would love to have this feature built

Plus one

I need one as well

plus one!

Just ran into this issue, I was designing ux as a separate project and building implementation in another one. I wanted to share the context from ux design into implementation so that I can implement the UX.

I think claude has all context across sessions/projects in the .claude folder. You just need a simple "@cross-refererence project/session" something like this will suffice.

(B/W - Most of this is to save token cost)

+1 on this. My specific use case: I manage multiple AI build projects and I split my time between Claude Code for implementation and claude.ai chat for planning, task management via Todoist MCP, and architecture discussions.

Every session I have to manually re-explain project context that Claude Code already has. There's no bridge between the two environments, so the context I build up in a coding session is completely invisible to my planning conversations and vice versa.

An MCP server that exposes Claude Code project context to claude.ai chat would solve this cleanly and fits the pattern Anthropic has already established with third-party MCP connectors. It would make the two products feel like one coherent workspace instead of two separate tools that happen to use the same model.

Closing for now — inactive for too long. Please open a new issue if this is still relevant.

This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.