[FEATURE] Allow Claude Code & Cowork to access Claude Chat Project knowledge bases

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

I use both Claude Chat (claude.ai) and Claude Code (CLI) for the same projects. In Claude Chat, I build up Project knowledge bases with context documents, system prompts, and reference material. In Claude Code, I maintain separate context via CLAUDE.md files, memory tools, and handoff documents, and other things like that.

These two context systems are completely isolated from each other. When I switch between tools (for example, starting an idea/ideating using Claude Chat and then perhaps switching to Claude Code for implementation) neither tool has access to the context I've built in the other. This means I end up duplicating effort, manually copying information between tools, and probably losing a lot of continuity. I'm sure Cowork would benefit from this too!

The core problem: context built in one Anthropic product is invisible to the other, even though they serve the same user working on the same projects sometimes.

Proposed Solution

Allow Claude Code and Cowork to read from Claude Chat Project knowledge bases, either via:

1. A new MCP server or CLI flag that authenticates with the user's Anthropic account and exposes Project knowledge bases as readable context;
2. A sync mechanism that exports Claude Chat Project files to a local directory that Claude Code and Cowork can read;
3. An API endpoint that Claude Code or Cowork can query to retrieve Project-level context.

Maybe the simplest way to do it would be claude --project "My Project Name" loads the associated knowledge base into the session context, the same way CLAUDE.md files are loaded today.

Alternative Solutions

Currently I manually maintain parallel context systems:

• CLAUDE.md files and memory tools in Claude Code;
• Project knowledge bases in Claude Chat.

I copy-paste a lot of key information between the two. For me to have any session continuity, I write handoff documents in Claude Code that I sometimes paste into Claude Chat projects and viceversa. It works, but it's manual, error-prone, and defeats the purpose of having persistent project context. I haven't even tried doing anything in Cowork yet, but I would imagine there could be great use in linking everything.

Priority

High - Significant impact on productivity

Feature Category

Other

Use Case Example

EXAMPLE:

1. In Claude Chat, I have a Project called "MicrasBot" for a worldbuilding Discord server. I have reference docs about my Discord bot's architecture, API specs, deployment procedures, as well as all the context needed for the worldbuilding setting and community the bot would operate in. The Claude Chat project knowledgebase has a lot of context about the Discord server, the lore developed, etc. that would be very helpful for Claude Code to perform its task well.
2. I switch to Claude Code to implement a new feature on MicrasBot.
3. Claude Code has no idea about all the context I've already built in Claude Chat about the bot, about Micras, about the worldbuilding, etc. - everything.
4. I have to either re-explain everything or maintain a duplicate CLAUDE.md with the same information.
5. With this feature, Claude Code could load the MicrasBot project context automatically and start with full awareness of Micras, of the bot's architecture, conventions, and decisions already documented in Claude Chat.

Additional Context

• Each product already has their own context systems (Chat: Project knowledge bases; Code: CLAUDE.md, memory MCP, handoff docs; etc.). The gap is that they don't talk to each other.
• Even read-only access from Claude Code and Cowork to Chat Projects would be a significant improvement.
• This would also help teams where some members use Chat and others use Code for the same project.