> 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/agent-context/integrations.md). # Integrations Understand how the Agent uses authorised external-system Connections and how web and channel run context affects which authority is available. The Agent can work with external business systems that are authorised for the current run. That can include email, ERP, CRM, spreadsheets, document stores and collaboration tools. A Connection gives Lleverage authority to call a provider; the Agent's instructions, Skills and run context determine how that authority is used for the operational task. ## Connections provide access, not process knowledge Connecting a system does not teach Lleverage how your operation works. For example, an ERP Connection may expose actions for reading or changing orders. A Skill can describe how your organisation validates an order, which sources to compare, which rules matter and when a person should be involved. **The Connection owns the credential and authority. The Skill owns the method.** ## Run context affects which Connections are available A Connection can be held in a Project, so the context of the run matters. A web Agent session opened inside a Project has an explicit Project context. Channel conversations can be routed differently. They may be pinned to a Project or, where the channel supports it, run without one fixed Project. In Microsoft Teams, an unpinned organisation-scoped conversation can resolve eligible project-held Connections only from Projects the user is authorised to access. That broader routing does not grant unrestricted access to other Projects. Pin the conversation when the work should be constrained to one Project's specific methods, context and Connections. If a tool works in one surface but not another, compare the run's Project/routing context and the Connection it resolved before reconnecting the external account. ## If the system is not connected yet For supported providers, the Agent can initiate connection setup when the task needs a system that is not yet authorised. The human completes the provider login or secure credential step. Do not paste passwords, API keys or other secrets into the conversation. Use the product's Connection, Secret or secure credential path. For the authority model and approval layers, see [Access and approvals](/run-operations/agent/agent-context/access-and-approvals.md). ## What you see during a session When the Agent uses an external-system tool, the session exposes tool activity around the work. Use that evidence to understand which systems were consulted or changed. For broader execution diagnosis, builders can inspect Monitoring. ## Authority still belongs to the external system The identity behind a Connection may have authority beyond the particular task. External-system permissions, mailbox delegation, tenant consent and service-account rights remain real even when Lleverage exposes only selected actions. Use approval controls and external permissions to create the right boundary for consequential writes. Do not treat a prompt instruction as a substitute for access control. If you administer Connections, see Connections and Data, connections and credentials. --- # 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/agent-context/integrations.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.