> 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/run-operations/agent/session-workspace.md). # Session workspace Understand the Agent session workspace: conversation, activity, files, context, memory, feedback and the controls around a task. A session is more than the chat transcript. It keeps one piece of work together with the controls, evidence and outputs around it. The exact controls can vary by Project access and enabled capabilities. ## Conversation and activity During a run, the centre stream can contain: * user and Agent messages; * collapsed activity rows you can expand to inspect work; * a run timer and stop control while work is live; * structured questions when the Agent needs input; * files, cards or other outputs; * the final result. Use the visible activity to understand the path the Agent took, but verify consequential outcomes against the system or artefact that owns the result. ## The composer The composer is where you continue the work. Depending on the current Project and capabilities, it can let you: * type the next instruction; * attach files; * use speech-to-text dictation to compose a message; * start an available Skill explicitly with **/skill** and choose the Skill you want; * choose from the model options available to you; * respond when the Agent is waiting for information. Dictation turns speech into text in the composer so you can review or edit the message before sending it. When the Agent is already configured for the job, prefer describing the business outcome over manually sequencing every tool it should call. ## In-session questions are not Requests The Agent can pause a session and ask a structured question when it needs operator input. A structured question can include selectable options, a free-text alternative, **Submit** and **Skip**. Choosing **Skip** declines to supply that input and lets the Agent continue with the context and judgement it still has. Treat that differently from a hard approval boundary. This is an **Agent session interaction**. It is not the same as a Workflow-created Request Decision or Request Input item in the Request Inbox. ## Files and inline preview Files produced by the Agent can appear in the conversation as attachment cards. You can open supported files in an inline preview pane and, depending on the file, use controls such as page navigation, zoom, download and save. A file being visible in the conversation does not automatically make it durable Project knowledge or a system of record. Saving or moving useful output into the appropriate maintained Project surface is a deliberate step. See [Working with files](/run-operations/agent/agent-context/file-handling.md). ## Progress, Files, Context and Memory The right-hand rail exposes supporting views around the current work. ### Progress Progress gives a compact view of a longer task's broad state. It is useful for following the run, but it is not proof that every downstream business action completed successfully. ### Files Files surfaces outputs and other files associated with the task. Important output should still be opened and inspected directly. ### Context Context helps show material and capabilities involved in the session. An empty panel should not be treated as proof that no tool or source contributed to the work. ### Memory Memory shows persistent context that may contribute beyond the current conversation. Remembered information is useful context, not automatically the authoritative business record. For the broader model, see [Memory](/run-operations/agent/agent-context/memory.md). ## Give feedback on a response or the whole session Completed Agent responses can show **How was this response?** with thumbs up, thumbs down and an optional comment. Use this when your judgement is about one specific answer. You can also leave persistent **session outcome feedback** when the judgement is about the piece of work as a whole, then return later to edit or remove your own feedback. ![The Agent session feedback dialog in production.](https://3370306531-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FlgSglJDQZxDZdn20CUcl%2Fuploads%2Fx1xLAXdsxQsFcrNsXJJY%2Fagent-session-feedback-dialog.png?alt=media) Feedback records human judgement of the result. It does not rewrite technical execution status. A technically successful run can still be a bad business outcome. Authorised builders can review response and session feedback alongside the execution evidence in [Monitoring and Activity](/build-and-improve/monitoring.md). ## Failed turns can be retried A persisted failed Agent response can expose **Retry**, letting you regenerate from the latest failed user message instead of reconstructing the conversation manually. Use retry for a failed turn. If the same underlying problem keeps recurring, fix the relevant Skill, Connection, context or execution path instead of normalising repeated retries. ## Use the workspace as evidence, not the system of record For important work, verify in this order: 1. the actual business-system result or required post-condition; 2. produced files, records and other concrete outputs; 3. visible Agent activity and the sources it used; 4. Monitoring or Activity evidence where available; 5. the session rail as supporting context. The workspace makes the Agent's work inspectable. It does not replace the systems that own operational truth. --- # 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/run-operations/agent/session-workspace.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.