> ## Documentation Index
> Fetch the complete documentation index at: https://docs.chamelio.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Agent Hub
> Build AI agents that take on a specific role for your legal team, start them from a template or from scratch, and track every run
Agent Hub is where you set up AI agents that each handle a specific role on your contract team, for example answering policy questions in Slack, watching a collection for upcoming obligations, or chasing unsigned envelopes. Each agent has a trigger that decides when it runs.
Open it from the **Agent** tab in the sidebar:
| Page | What it's for |
| ---------------- | -------------------------------------------------------------------------------------------------------- |
| **Agents** | The agents you and your team have built, in three tabs: **My Agents**, **Shared Agents**, and **Drafts** |
| **Library** | Ready-made agents you can start from |
| **Activity log** | Agent runs, with their status, trigger, and results |
## Start From the Library
The **Library** ("Agent library") lists ready-made agents, each built for a role on a contract team. Search by role, task, or keyword, or filter by category: **Contract Review**, **Negotiation**, **Legal Research**, **Compliance**, **Notifications**, and **Data Extraction**.
Examples from the Library:
| Agent | What it does |
| ---------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **The Paralegal** | Reads every incoming contract, summarizes it, and flags risky clauses against your playbook |
| **The Negotiator** | Drafts redlines on counter-party paper using your fallback positions and past deal patterns |
| **Renewal Sentinel** | Each week, finds contracts expiring in the next 30 days with no auto-renewal clause and drafts a termination notice and a one-year extension addendum for each |
| **Contract Obligation Watchtower** | Scans a contract collection monthly for obligations due within 30 days or overdue, and posts a digest to Slack |
| **Signature Chaser** | Sends documents for signature with Chamelio Sign and chases envelopes that stay unsigned for more than 2 days |
Click an agent to see its details before you use it:
* **Category**, **Tools used**, **Owner**, and **Agent mode**
* **Why use this agent?** - what it's for
* **Definition of done** - what a finished run looks like
* **Required resources** - what you need to connect, such as a Slack channel or a playbook
Click **Use this agent** to start from it, then customize it.
## Build an Agent From Scratch
On the **Agents** page, click **New agent**. An agent's configuration has three steps.
Give the agent a name and an optional description, then choose when and how it starts:
* **Manual**
* **Scheduled**
* **Slack** - choose the **Message type** (for example, **Mention agent**) and the **Channels**
* **Contract activity**
Write the **Instruction**: what the agent should do. Type `@` to point it at specific resources, such as a Vault folder. You can also add a **Definition of done** (the conditions for a successful run) and **Expected outputs** (what the agent produces when it finishes). Both are optional.
Choose how the agent works and what it can use:
* **Model Type** - **Fast**, **Focus**, or **Plan**
* **Escalation** - by default, anything that needs human review is sent to you. Use **Customize human review assignee** to send it to someone else.
* **Tools** - the tool groups the agent can use, each with its own permission setting. For example, **Communication** can find the right person or Slack channel and send Slack, email, or Teams messages after you approve them.
Use **Test agent** to try it, then **Publish** it.
## Manage an Agent
Open any agent to see four tabs:
| Tab | What you'll find |
| ----------------- | ------------------------------------------------------------------------------------------------------------------------------------------------ |
| **Overview** | Total runs, the trigger, the last edit, the resources it uses, the last run, and a **Runs over time** chart |
| **Runs** | Each run with its status, run type, start time, trigger, duration, and last interaction |
| **Configuration** | The Trigger, Task, and Capabilities steps above |
| **Memory** | What the agent remembers, in **General**, **Tool strategy**, and **Outputs**. Type in the box to tell the agent something to remember or forget. |
From the agent's page you can also:
* Switch between saved versions with the version selector (for example, **Version 4 (Active)**)
* **Pause** an active agent
* Change it and **Publish** again
## Track Runs Across Agents
The **Activity log** lists agent runs so you can monitor their status, trigger, and results. For each run you'll see the agent name, status (for example, **Completed** or **Failed**), run type, trigger (such as **Manual** or **Slack**), run number, start time, and duration. Use **Search** and **Filters** to find a specific run.
## Related
Chat with the Agent directly for reviews, research, and drafting
Automate multi-step requests with intake, approvals, and signatures