> ## Documentation Index > Fetch the complete documentation index at: https://docs-agents.fpt.ai/llms.txt > Use this file to discover all available pages before exploring further. # Agent details > The parts an agent is made of, and what each one decides Clicking an agent in the list opens its own page. On the **Agent builder** tab, the left column lists the parts that make up the agent, while the right column gives you a quick read on how it is currently configured. Agent details ## The parts of an agent | Component | What it decides | | - | - | | [**Instructions**](/en/instructions) | Who the agent is, what it can do, in what order it works and how it speaks | | [**Model**](/en/model) | The brain behind the agent - reasoning ability, context length and cost | | [**Skills**](/en/skills) | Repeatable jobs packaged up for the agent to reach for at the right moment | | [**Connections**](/en/connections) | Which systems the agent may read from and write to, and under whose account | | [**Guardrails**](/en/guardrails) | What the agent must not do, and which actions need a human to approve | Each part has its own page in this section. ## The publish readiness bar At the bottom of the left column, a bar tracks whether the agent has everything it needs: Instructions written, a model chosen, tools attached, guardrails configured and a test run completed. Treat it as a suggested route, not a hard requirement. **Knowledge**, **Triggers** and **Sub-agents** are all marked *Coming soon* for now - they will be documented once they go live. ## Conversation history of an agent History now belongs to the Agent itself. An Agent shared on a Workspace keeps every conversation under that Agent, so you can read them all back no matter which Workspace the Agent was opened from. Open the Agent, go to the **Analytics** tab and click **History**. Conversation history of an agent The list has these columns: | Column | Meaning | | - | - | | **End time** | When the conversation ended | | **Conversation ID** | The identifier used to search and to raise a support ticket | | **Channel** | Where the conversation came from: Workspace, Test, Web, API, Zalo, Teams or Scheduled | | **User** | Who was talking to the Agent | | **Messages** | How many messages the thread holds | Above the list, **All channels** narrows the view to one channel and **All time** sets the period. The search box takes message content, a sender name, a conversation ID or a message ID. ### Reading one conversation Click any row and **Conversation details** opens on the right with the whole thread in order. Steps where the Agent called a tool appear as **Tool call** entries, so you can see what it looked up before it answered. Conversation details History is read only: This screen does not edit or delete conversations. **Performance**, the other item on the Analytics tab, is still in development and will arrive later. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.