> For the complete documentation index, see [llms.txt](https://docs.talus.network/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.talus.network/concepts/01-talus-agent.md). # The Talus Agent {% hint style="info" %} **Audience:** Agent developers and application architects. **Goal:** Understand the Agent identity, skill contract, custody, and payment boundary before authoring a workflow. {% endhint %} An Agent is an on-chain identity that groups one or more skills and is created with a standard `AgentPaymentVault` child. Using that vault to fund executions is optional: an Agent-funded skill may draw from it, while a user-funded skill uses the selected user or address source. A skill is the public contract for a workflow: it binds a DAG to input, payment, schedule, fixed-Tool, and interface-revision requirements.
Agent identity and runtime boundaries
The Agent object is the custody handle for its identity and vault. The registry record makes the skill contract discoverable; it does not replace custody of the Agent or the application state associated with it. ### Skills package intent A skill records the DAG a run should execute, the committed input shape, the accepted payment source, the allowed schedule shape, fixed Tool dependencies, and the active interface revision. New Tasks pin the skill contract they use so a later skill update does not silently change an in-flight Task. An Agent can be created and managed directly, or embedded inside an application shared object. The embedded pattern is useful when a TAP wants one package-defined object to hold business state and the Agent identity together. Either pattern still requires explicit custody of the Agent and each application capability. #### Default and application Agents The protocol can expose a default Agent for low-ceremony DAG execution, while application developers normally create an Agent whose skill records describe their own workflow. The default path does not make application business state or Tool authority implicit; those remain the responsibility of the caller and Tool package. #### Payment and state boundaries An Agent-funded execution draws from the Agent payment vault; a user-funded execution draws from the user or application-selected execution source. Payment state and runtime state are explained in [Payment Vaults, Reserves, and Settlement](/concepts/06-payment-vaults-reserves-and-settlement.md) and [DAG and Execution](/concepts/05-workflow-dag-execution.md). Durable business objects belong in [Assets and Business State](/concepts/02-assets-and-business-state.md). #### Agent design checklist 1. Define the Agent identity and who retains mutable custody. 2. Publish or select the DAG that each skill should bind. 3. Commit input, payment, schedule, fixed-Tool, and interface-revision requirements. 4. Keep business-state capabilities outside generic execution state and document their mutation invariants. 5. Verify the Agent registry record and the Agent object after each registration or skill update. **Next** Continue to [Assets and Business State](/concepts/02-assets-and-business-state.md) to design what the Agent owns or uses. --- # 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 following URL with the `ask` and `goal` query parameters: ``` GET https://docs.talus.network/concepts/01-talus-agent.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `build a script that syncs our docs to a CMS` lets GitBook tailor the answer to that use case. 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.