> ## Documentation Index
> Fetch the complete documentation index at: https://docs.rippit.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Ada

> Let the Rippit agent read how your Ada AI Agent performs and how it is configured, alongside your conversation data.

Connect Ada when questions about your customer conversations lead back to your Ada AI Agent: why containment dropped for a topic, which knowledge article the agent leaned on in a bad answer, or what changed in its configuration before CSAT moved. The Rippit agent reads Ada's metrics, conversations, knowledge, and configuration and cites what it finds in its answer.

## What the agent gets

* **Metrics**: automated resolution rate, CSAT, containment, and engaged conversation volume, with conversation-level metadata.
* **Conversations**: filtered conversation lists and full transcripts.
* **Knowledge and coaching**: search across your knowledge articles and coaching entries.
* **Configuration**: Playbooks, tools, custom instructions, topics, variables, handoffs, and the other entities your Ada AI Agent is built from.
* **Change sets and tests**: staged and promoted change sets, metric comparisons against baseline, test cases, and test run results.

## Setup

Nothing to configure in Ada. Rippit registers itself as an OAuth client automatically, so a Full Admin only supplies the URL and signs in.

1. Find your Ada instance domain: the address you use to log in to the Ada dashboard, such as `acme-corp.ada.support` or `acme-corp.eu.ada.support`.
2. Add `/api/mcp/oauth` to it. Your MCP server URL looks like `https://acme-corp.ada.support/api/mcp/oauth`.
3. In Rippit, go to **Connections → MCP Servers**, choose **Ada**, and paste the URL. Rippit accepts only an `ada.support` host ending in `/api/mcp/oauth`. Ada's `/api/mcp` URL is for API keys and cannot complete Rippit's sign-in.
4. Click connect and log in with your Ada account.

Ada's own guide is at [Ada MCP authentication](https://docs.ada.cx/mcp/introduction/authentication).

## Identity and scope

The connection acts as the Ada user who authorized it, and your Ada dashboard role decides which tools appear. Owners and Admins see every tool, including the ones that make changes. Users with the Agent or Read Only role see the read tools only, and a role change in Ada applies on the next tool call without reconnecting.

An organization-wide connection shares that Ada user's access with everyone using the agent. Authorize with an account whose access you are comfortable extending that far.

The **scope** you choose at setup decides whether Agent Apps can use the connection, exactly as with every other connector.

## Tools

Ada marks its read tools as read-only, and those start enabled. **Manage tools** on the connection card lists every tool your connection has.

| Tool | What it does |
| - | - |
| `get_ada_metric` | Read resolution rate, CSAT, containment, and conversation volume |
| `get_available_filters` | List the filters conversation queries accept |
| `get_conversations` / `get_conversation` | List conversations and read a full transcript |
| `search_knowledge` / `search_coaching` | Search knowledge articles and coaching entries |
| `get_ada_configuration` | Summarize Playbooks, tools, custom instructions, and company description |
| `list_entities` | List channels, topics, variables, Playbooks, handoffs, knowledge sources, and other configured entities |
| `get_improvement_guide` | Read Ada's guide to entity types and how to propose improvements |
| `list_glossary_terms` / `get_glossary_conflict_guide` | Read glossary terms and Ada's steps for finding where the rest of the configuration contradicts them |
| `list_agent_changesets` / `compare_changeset_metrics` | Review change sets and compare their metrics against baseline |
| `get_test_cases` / `get_test_runs` / `get_test_run_quota` | Read test cases, test run results, and the remaining test run quota |
| `send_feedback` | Send feedback about Ada's MCP tools to Ada's development team. It does not change your AI Agent |

Tools that Ada does not mark read-only start off. They include `edit_agent_behavior`, which stages edits to knowledge, coaching, Playbooks, and custom instructions, then promotes or reverts them, and `edit_agent_config`, which changes settings such as topics, test cases, and scorecards, triggers test runs, and applies immediately. Turn a tool on in **Manage tools** only if you want the agent to take that action.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.