> 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/chat-and-sessions.md).

# Sessions and continuity

Keep each Agent session focused on one coherent outcome while durable Project context carries useful knowledge across sessions.

A **session** is one coherent piece of work with the Agent.

It keeps the conversation together with the surrounding evidence: activity, files, tool/context use and relevant memory.

## Same outcome: keep going

Continue in the same session while you are still working toward the same result.

That includes:

* answering a clarification;
* attaching another file for the same task;
* correcting a conclusion;
* inspecting and revising an output;
* continuing a multi-step investigation.

The earlier work remains part of the context for the next turn.

## New outcome: start a new session

Start a new session when the job genuinely changes.

For example, finishing a supplier-delivery investigation and then asking for a quarterly inventory policy review are two different pieces of work, even if both involve purchasing.

Keeping sessions scoped makes the active context easier to reason about and the later history easier to inspect.

## The organisation's knowledge does not live only in one transcript

Do not keep one endlessly growing session alive just to preserve how your organisation works.

Durable context belongs in the appropriate Project layer:

* recurring operating method → Skill;
* exact maintained data → source system or Table;
* approved reference material → Knowledge;
* useful persistent working context → Memory;
* repeatable controlled sequence → Workflow.

A new session can still benefit from those maintained capabilities.

## Give enough context for the current job

A useful request normally makes clear:

* **Outcome** — what needs to be done.
* **Relevant material** — files, systems, Skills or other evidence.
* **Constraints** — what may or may not happen.
* **Expected result** — what you need back to judge the work.

You do not need to restate the whole Project setup on every turn.

## The transcript is only one part of the session

Use [Session workspace](/run-operations/agent/session-workspace.md) to inspect Progress, Files, Context and Memory alongside the conversation.

For important work, the strongest proof is still the actual post-condition in the business system or output, not a confident final sentence.

## Questions and approvals stay with the work

When the Agent asks for information or approval that belongs to the current job, answer in the same session.

That is different from a Workflow-created Request Inbox item. A question inside Agent chat does not by itself create a Request.

## Slack and Teams conversations

Where configured, Agent work can also start in Slack or Microsoft Teams.

A channel conversation can be pinned to one Project or, where supported, run against the acting user's authorised Project scope. When the work clearly belongs to one Project, pinning gives the Agent an unambiguous capability/context boundary and makes later inspection easier.

See [Agent in Slack](/run-operations/agent/agent-in-slack.md) and [Agent in Microsoft Teams](/run-operations/agent/agent-in-microsoft-teams.md).

## When a correction should survive the session

If a correction changes the durable way the work should be done, move it to the layer that owns it instead of trusting everyone to repeat the correction in future chats.

That is how sessions stay useful as workspaces rather than becoming hidden configuration.


---

# 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/chat-and-sessions.md?ask=<question>&goal=<endgoal>
```

`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.