Use Cursor 3.11 Side Chats Before Your Main Agent Loses Context
Cursor 3.11 adds side chats, agent transcript search, and new cloud-agent hooks, giving developers a cleaner way to manage long-running coding agents without interrupting the main task.
Cursor 3.11 is a confirmed release for developers managing long-running coding agents. Confidence level: confirmed. The update adds side chats, agent transcript search, simplified repo pickers, and cloud-agent hooks that make it easier to ask questions or inspect past work without derailing the main agent thread.
What changed
Cursor now lets users open side chats with /side, /btw, or the plus button in the chat panel. A side chat inherits context from the main conversation, stays durable, and can be referenced back into the main thread later. That makes it useful for research, clarification, and second-pass checks while the main agent keeps working.
The release also adds local agent transcript search from the Agents Window and in-chat search with Cmd+F. Cursor says it builds a local search index that can handle thousands of conversations. For teams using many agents, this changes old agent chats from disposable logs into reusable workflow memory.
| Feature | Best use | Access | Caveat |
|---|---|---|---|
| Side chats | Ask questions or test alternatives without stopping the main agent | Cursor 3.11 | Keep side-chat conclusions explicit before pulling them into production work |
| Transcript search | Find old agent runs, PR context, and decisions | Agents Window and in-chat search | Local index quality depends on what was captured |
| Project and repo pickers | Pick local, cloud, and remote work targets faster | Cursor 3.11 | Verify the selected machine or repo before launching agents |
| Cloud-agent hooks | Observe prompts, responses, thoughts, subagents, compaction, and turn completion | Cursor cloud-agent workflows | Hooks can add control, but they also need review and testing |
Why this is early
Cursor published the changelog on July 10, 2026. AlphaSignal covered the update as a shift toward parallel agent conversations, and Releasebot's Cursor timeline corroborates the side-chat, transcript-search, repo-picker, and cloud-agent-hook details.
This is not a frontier model launch, but it is a meaningful developer-tool update. The bottleneck in coding-agent work is often attention and context management, not only model intelligence. Cursor 3.11 targets that bottleneck directly.
Key takeaways
- Side chats let developers investigate a tangent without interrupting a running agent.
- Agent transcript search makes prior Cursor sessions easier to reuse.
- New cloud-agent hooks expose more of the agent lifecycle for observation and control.
- Repo and project picker changes reduce launch friction across local, cloud, and remote machines.
- Teams should still validate side-chat conclusions before merging them into the main workflow.
Availability and access
Cursor lists the update under version 3.11 with a July 10, 2026 changelog date. Users should update Cursor and check whether side chats, transcript search, and the revised pickers are visible in their workspace. Team and cloud-agent features may depend on the plan, workspace setup, and admin configuration.
The cloud-agent hook changes are most relevant for teams already using Cursor's asynchronous agents. Before relying on hooks for guardrails or observability, test them on a low-risk repo, confirm when each hook fires, and document who owns failures.
Practical LinkLoot angle
Use Cursor 3.11 to separate three kinds of work: the main implementation thread, quick side investigations, and searchable post-run evidence. That split helps teams avoid stuffing every question, detour, and memory into one fragile agent conversation.
For agent-heavy teams, the first useful workflow is simple: start a main coding agent, open a side chat for risk checks or API questions, then pull only the relevant result back into the main thread. Pair that with LinkLoot's AI agent tools guide when comparing Cursor with Codex, Claude Code, GitHub Copilot, and local agent stacks.
What to verify before you act
- Confirm Cursor 3.11 is installed and the side-chat controls appear in your chat panel.
- Test whether side chats preserve enough main-thread context for your codebase.
- Search old agent transcripts for a known PR, issue, or decision before relying on the index.
- Review cloud-agent hooks in a sandbox repo before using them for production controls.
- Check team settings for remote machines, cloud agents, and connected Git providers.
Source check
Confirmed by:
- Cursor's official changelog, which names side chats, agent transcript search, repo-picker changes, and cloud-agent hooks.
- Releasebot's Cursor timeline, which independently tracks the July 10 update and summarizes the same feature set.
Early signal / context:
- AlphaSignal frames the update around parallel agent conversations and the attention cost of managing long-running coding agents. LinkLoot treats that as workflow context, not as a separate product claim.
Cursor 3.11 adds side chats, agent transcript search, simpler project and repo pickers, and cloud-agent hooks for observing more of the agent lifecycle.