[FEATURE] Method to Cross-link specific Claude Code sessions with claude.ai chat sessions that need content reference
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
Summary
There's currently no way to reference or pull content from a Claude Code session into a claude.ai chat session (or vice versa). When work starts in Claude Code (terminal diagnostics, infra config, file edits) and later needs to be discussed, summarized, or handed off in a claude.ai chat — or when a chat-side decision needs to inform an active Code session — the only options are manual copy-paste (token costly), requesting Claude generate a session summary (lossy), or losing context entirely.
Problem
- Claude Code sessions are stored locally (
~/.claude/projects/) and tied to project path + machine; they aren't visible or searchable from claude.ai. - claude.ai's
conversation_search/ memory tools only index claude.ai chat history, not Code sessions. - Users working across both surfaces (e.g., infra work in Code, planning/tracking in chat) have to manually restate Code session context in chat, or export/paste raw transcripts.
- This gets worse across multiple machines/accounts, where a Code session may not even be locally accessible to the person now working in chat.
Proposed Solution
Requested feature
A lightweight link/reference mechanism so a claude.ai chat can pull relevant context from a named Code session (or vice versa) — e.g.:
- A way to tag or name a Code session so it can be referenced by ID/name from a chat
- A chat-side tool (similar to
conversation_search) that can search/retrieve Code session transcripts or summaries, scoped to sessions the user has access to - At minimum, a supported "handoff" export format from Code that a chat session could ingest as structured context rather than a flat text transcript
- If a handoff export of each session could be uploaded periodically to account storage, reference by name or code-session id (which could be searched as a lookup) could happen from any surface the user needs.
Ideally, this process should work while using Chat in Claude Desktop to reference local Claude Code (in Claude Desktop) sessions. Then the further reach of the feature can be added incrementally. For my specific case, I am most likely to need this in the Windows version of Claude Desktop. But doing this properly could allow for cross-reference in a mobile app chat about a Code work session that was done in terminal days ago.
Alternative Solutions
My current prompt for referencing Chat sessions in other contexts isCan you summarize this conversation in a way that's useful for a different agent session to pick it up and incorporate it into work it's currently doing?
That works well for carrying a chat summary to other places, but not well for varieties of code and repo layouts (that may or may not have formal decision structures) going in the direction of chat sessions.
Priority
High - Significant impact on productivity
Feature Category
Other
Use Case Example
Solo IT/infra admin runs long Claude Code sessions doing hands-on deployment work (AWS/infra config, diagnostics), then switches to claude.ai chat for planning, prioritization, and non-technical drafting. Needing to reference "what did the Code session already do/decide" from chat currently requires manual re-summarization.
Additional Context
_No response_
3 Comments
+1 on this — I ran into the same disconnect from a slightly different angle: the "brainstorm in chat, then implement in Code" workflow.
A common pattern for me is: use claude.ai chat to think through / design something, and once it's solid, switch to Claude Code to actually build it. Days later I often can't remember whether a particular decision was made in chat or in a previous Code session, and there's currently no way to search across both.
A couple of concrete ideas that would help, in addition to the cross-linking mechanism proposed above:
/search-chat <keyword>that queries the same account's claude.ai chat history and surfaces relevant past conversationsThe reviewer with the most context is the only participant locked out
Workflow: one claude.ai Project chat as standing reviewer; multiple named
Claude Code sessions plus plain shell tabs (git, file ops, local model
runs) as workers in a git repo.
The chat is the natural supervisor of both:
project memory and prior decisions. It reviews every worker report,
drafts every worker instruction, and dictates the exact shell commands
I run.
models, release notes, and Code capabilities. It told me a new model
shipped and that inter-session messaging existed. Workers don't know
their own tooling.
But it can't see any of it. Every crossing is manual: drag report files
into chat, paste instructions into terminals, paste command output back,
screenshot scrollback to answer "is it done, did it stall" — lossy samples
of what are already text streams, dozens of times a day.
SendMessage/ListAgents proved the concept worker-to-worker. The request,
opt-in per session/tab:
Claude Code session or plain shell
the surface boundary)
Opt-in grants preserve the one virtue of the manual ferry — nothing crosses
without user review. Everything else about it is tax.
+1 on this.
My specific pain point: I work on the same systems across dozens of Claude Code
sessions over months. Each session has context about architecture, deployment,
past bugs — but none of it carries into new sessions or into claude.ai Projects.
claude.ai now supports importing chat conversations into Projects, which proves
the infrastructure exists. The most useful next step for me would be:
The knowledge is already there in session history — it just needs a way out.