Feature request: shared context between Claude Code, Claude.ai and Cowork

+1. This siloing between apps is counterintuitive to me. When I designate a "project" on claude.ai, I want that project to have access to all y work on the subject, including things where I want to execute code iteratively, which, as of now, only claude code can do. I'm sure there are workarounds (I could ask Claude Code to make a change log and upload that to the web app) but i'm surprised by the lack of built-in functionality for this pretty obvious use case.

shared context between Claude Code and Claude.ai would be huge. right now we maintain a file-based memory system that only Claude Code can read, so any insights or decisions made in a Claude.ai conversation are lost. even just syncing the CLAUDE.md and memory files bidirectionally would close the biggest gap. for Cowork specifically, being able to share the project context (build commands, testing workflow, architecture decisions) would save a lot of re-explaining.

here's the file-based memory system we maintain in Claude Code that we wish could sync to Claude.ai: https://github.com/m13v/fazm/blob/main/CLAUDE.md

Exact same experience here. We've built a fairly extensive file-based memory system that Claude Code maintains across sessions — separate topic files for context, preferences, feedback, and project state, plus a session log for continuity. Combined with MCP integrations (calendar, email, WhatsApp, CRM, task managers, etc.) it effectively turns Claude Code into a persistent working environment rather than a one-off conversation.

The gap is real though: anything discussed in claude.ai stays there. We work around it by using Drafts (iOS/macOS) as an intake layer — dictated notes, ideas, rough text get tagged and processed by Claude Code in the next session. But it's a workaround, not a solution.

Bidirectional sync of memory and project context between Claude Code and claude.ai would be transformative.

I would like to upvote this. The siloing of context within Cowork when I have a well tuned claude code with a lot of memories, skills, hooks, etc is quite frustrating.

Filed a sharper-scoped framing of this at #55842 — proposes unified user state across Cowork and chat with four pillars (memory, files, skills/connectors, surface handoff). The Cowork ↔ mobile-chat axis specifically isn't well covered anywhere; that issue tries to articulate it explicitly.

I wanted this badly enough that I built an open-source workaround: Thruline — one shared memory vault across Code, Cowork, and Chat.

Claude Code connects over stdio (local JSON or Supabase)
Cowork and Chat reach the same vault through a custom connector (Supabase Edge Function over HTTPS)
Sessions are resumable snapshots: save in Code, open Cowork, say "continue where we left off" — the context is already there. Every entry records which surface it came from.
It's early and the limits are documented honestly in the README (single-user, no auth on the remote path yet), but the cross-surface handoff works today and I use it daily. Sharing in case it helps anyone here while we wait for a native solution.

Same wall here. We ended up building BaseThread: an MCP server that Claude Code, Cursor, and ChatGPT all read and write the same team context to, so skills, decisions, and project state carry over across whichever tool you're in, not just within Claude Code. Doesn't fix the Cowork/claude.ai side specifically, but solves the cross-tool half of this for a team. https://basethread.ai

Strongly behind this. The one surface I'd explicitly fold into the unified state is voice. Right now voice is the most fragmented of all. My Cowork and Claude Code CLI sessions carry my whole second brain, my vault, my memory, my accumulated context, but the moment I open a voice conversation, in Cowork or on mobile, all of that vanishes and I'm talking to a stranger. If context handoff becomes seamless across surfaces, voice has to be a first-class citizen in that set, not an afterthought. Being able to say, continue this in voice, and have my full context come with me, would be the single biggest unlock. I don't want to be the glue anymore, especially not when my hands are busy and voice is the only way I can work.

I would love to be able to seamlessly connect claude.ai, claude design, and claude code into a pipeline that attempts to cover idea to design to code. I am currently working on a project where I have designated claude.ai using project instructions as a "Collaborator" / "Business Analyst" / "Executive Assistant" role, responsible for collaborating with me on business strategy, positioning, marketing, etc. Claude Code as the "Coder" / "CTO" role that is responsible for grooming and implementing issues, maintaining github actions workflows, and other operations activities such as managing IaC, and Claude Design as the "Designer" role which iterates with me on ticket created by the collaborator during ideation into design specs to hand off the claude code. The main friction right now is integrating the actors in the system. Currently I have to ask for messages to copy paste or ask to update the github issue with notes which is annoying.