Feature request: Bidirectional links between Chats, Plans, and Canvases

Feature request for product/service

Cursor IDE

Describe the request

Feature request: Bidirectional links between Chats, Plans, and Canvases

Product: Cursor (Agent / Plan mode / Canvas)
Submitted by: (your name)
Date: 2026-09-21

Problem

When I close a chat tab, I lose the obvious path back to artifacts that conversation created or updated:

Plans (e.g. ~/.cursor/plans/*.plan.md or Plan-mode output)

Canvases (in-chat or standalone analytical artifacts)

Related outputs (gists, PRs, docs) are also disconnected unless I manually bookmark them

Reverse lookup is equally broken: opening a plan file or canvas does not show which chat produced it or offer “continue this thread.”

Today recovery is manual archaeology (filesystem search, gist list, chat history keyword search, agent transcripts). That does not match the mental model: the chat is the session; plans and canvases are deliverables of that session.

Requested behavior

  1. Chat → artifacts (persistent)

Each chat (including after the tab is closed) should show a Linked artifacts section, automatically populated when the agent:

Creates or updates a Plan

Creates or updates a Canvas

(Optional) Creates/updates other first-class artifacts the product knows about (e.g. published gist, automation draft)

Each entry: title, type icon, path or deep link, last modified time, one-click open.

  1. Artifact → chat (reverse link)

From a Plan or Canvas UI (and from the file metadata where applicable):

Source chat — title + link to reopen or fork the conversation

Continue in chat — resume with artifact context loaded

  1. History and search

Chat history list/search should include artifact titles (not only message text).

Global search: find plans/canvases and jump to source chat.

  1. Consistency across modes

Plan mode, Agent mode, and Canvas should share the same linkage model so switching modes does not orphan artifacts.

Why it matters

Long-running infra work (multi-repo, multi-PR) depends on plans and canvases as the durable spec; chats are ephemeral UI.

Teams cannot rely on “remember which tab” or repo-local conventions alone.

Reduces duplicate work (re-explaining context) and wrong-thread edits.

Non-goals / nice-to-have

Full knowledge base across all repos (not required for v1).

Auto-sync to external systems (Jira, Slack) — optional later.

Success criteria (v1)

Close chat tab → reopen from history → Linked artifacts still visible and clickable.

Open a canvas created in a chat → see Source chat and reopen it.

No manual path pasting required for default agent-created plans/canvases in that session.