> For the complete documentation index, see [llms.txt](https://docs.lleverage.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.lleverage.ai/operator-agent/chat-and-sessions.md). # Chat & Sessions Every conversation with the Operator Agent is a **session** (also called a thread). A session preserves the full context of that conversation - messages, uploaded files, generated outputs, and the agent's working notes. ### How Sessions Start Sessions don't only begin when you type a message. A session can start from: * **Chat** - You open the agent and send a message. * **A schedule** - The agent runs a task on a recurring schedule and records the run as a session. * **A messaging channel** - Messages from connected channels such as Slack or Microsoft Teams start and continue sessions. * **A workflow** - Workflows can hand tasks to the agent, which run as sessions too. You can start a new session at any time from the agent home screen, continue an existing one from the session list, and run multiple sessions in parallel for different topics. ### The Composer The message box does more than plain text: * **Mentions** - Type `@` to reference a workflow or data table, or `/` to reference a skill, directly in your message. The agent then knows exactly which resource you mean. * **The + menu** - Attach resources to the conversation from four panels: Skill, Workflow, Data table, and Connections. * **Voice input** - Dictate your message instead of typing it. * **Queued messages** - Send follow-ups while the agent is still working; they queue up and are handled in turn. You can remove a queued message before it's picked up. * **Stop** - Stop the agent mid-response if it's heading the wrong way, then redirect it. ### Choosing a Model A settings control in the chat lets you choose how the agent thinks for the current session: * **Auto** - Lets the platform route each request to a suitable model automatically. * **A specific model** - Pick an exact model if you have a preference. * **Thinking** - For models that support it, choose how much reasoning effort the agent puts in. > 💡 Model settings apply per session, so you can use a fast model for quick questions and a more thorough one for complex work. ### The Right Panel Alongside the chat, a panel keeps track of the session: * **Progress** - The agent's todo list for longer tasks, ticked off as it works. * **Files** - Files generated or used during the session. * **Context** - The skills, connections, and other resources the agent is drawing on. Selecting a skill opens it for review and editing without leaving the chat. * **Memory** - What the agent has learned and remembered for this project. ### Requests from the Agent When the agent needs something from you mid-task - approval to proceed, an answer to a question, or a file - it raises a **request**. The request appears as a card in the session, and also on your project's Requests page, so nothing gets missed even if you've moved on. Respond from either place and the agent carries on. ### Chatting from Slack and Teams You can connect messaging channels so the agent is available where your team already works. Once a channel is connected, you can chat with the agent directly inside Slack or Microsoft Teams, and those conversations appear as sessions in Lleverage. ### Managing Sessions The sidebar shows your recent sessions for quick access. For the full picture, open the **Agent Sessions** page, where you can: * **Search and filter** sessions - for example by status, such as those needing attention, active, or completed * **Favourite** sessions you want to keep close to hand * **Rename** sessions so they're easy to find later * **Archive and unarchive** sessions to keep the list tidy without losing anything * **Delete** sessions you no longer need, individually or in bulk ### Session Privacy Sessions are scoped to your user account. Other users in the same project cannot see your sessions, and you cannot see theirs. ### Context Within and Across Sessions Within a session, the agent remembers everything - previous messages, file uploads, skill results, and generated outputs. You can refer back naturally: * *"Use the same format as the report you generated earlier"* * *"Now do the same thing for the second file I uploaded"* * *"Actually, change the date column to DD/MM/YYYY"* For very long sessions, the agent manages its context window automatically to stay within model limits. Across sessions, the agent isn't starting from scratch either: project instructions, tone, and the project's memory carry over, so the agent keeps what it has learned about your team and how you like to work. Open a new session when you want a clean conversation for a new topic - the conversation itself starts fresh, but the agent's knowledge of your project comes with it. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.lleverage.ai/operator-agent/chat-and-sessions.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.