> For the complete documentation index, see [llms.txt](https://doc.wearepatchworks.com/product-documentation/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://doc.wearepatchworks.com/product-documentation/developer-hub/patchworks-mcp/patchworks-mcp-hosted/default-tools-for-the-hosted-patchworks-mcp-server.md).

# Default tools for the hosted Patchworks MCP server

## Introduction

The hosted Patchworks MCP server includes a default set of tools that are ready to use. These tools are summarised here, in the following categories:

* [Flow management](#flow-management)
* [Data pools](#data-pools)
* [Marketplace](#marketplace)
* [Troubleshooting](#troubleshooting)
* [Utility](#utility)
* [Agent conversation - platform agents](#agent-conversations)

## Flow management

<table><thead><tr><th width="302.65625">Tool</th><th>Summary</th></tr></thead><tbody><tr><td>get_all_flows</td><td>List flows from the Core API.</td></tr><tr><td>get_flow_runs</td><td>Query flow runs (filter by status; started_after; sort; includes).</td></tr><tr><td>get_flow_run_logs</td><td>Retrieve logs for a specific flow run (optionally with payload IDs).</td></tr><tr><td>start_flow</td><td>Trigger a flow run via the Start service (/flows/{id}/start).</td></tr></tbody></table>

## Data pools

<table><thead><tr><th width="311.86328125">Tool</th><th>Summary</th></tr></thead><tbody><tr><td>list_data_pools</td><td>List all data (dedupe) pools.</td></tr><tr><td>get_deduped_data</td><td>Retrieve deduplicated data for a specific pool.</td></tr></tbody></table>

## Marketplace

<table><thead><tr><th width="312.76953125">Tool</th><th>Summary</th></tr></thead><tbody><tr><td>get_marketplace_apps</td><td>List marketplace blueprints from the Patchworks marketplace.</td></tr><tr><td>get_marketplace_app</td><td>Get details of a specific marketplace blueprint by ID.</td></tr></tbody></table>

## Troubleshooting

<table><thead><tr><th width="307.2421875">Tool</th><th>Summary</th></tr></thead><tbody><tr><td>summarise_failed_run</td><td>Summarise what went wrong in a failed run.</td></tr><tr><td>triage_latest_failures</td><td>Fetch recent failed runs and return a compact summary.</td></tr><tr><td>download_payload</td><td>Download payload bytes for a given payload metadata ID (base64).</td></tr></tbody></table>

## Utility

<table><thead><tr><th width="307.2421875">Tool</th><th>Summary</th></tr></thead><tbody><tr><td>example_get_all_flows</td><td>Example demonstrating tool definition patterns.</td></tr><tr><td>list_tools</td><td>List all available tools in this MCP server.</td></tr></tbody></table>

## Agent conversations <a href="#agent-conversations" id="agent-conversations"></a>

These tools provide low-level access to the Patchworks agent conversation API. They can be used to start and continue a conversation with any platform agent directly. For purpose-built wrappers with clearer parameter guidance, see Platform agents below.

| Tool                          | Summary                                                                                                                                                                                                    |
| ----------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `list_agent_conversations`    | List all agent conversations (paginated).                                                                                                                                                                  |
| `create_agent_conversation`   | Create a new agent conversation for a given feature (`assistant`, `implementation-agent`, `connector-builder`, `flow-builder`, `map-builder`, `answers-agent`, `documentation-agent` or `health-monitor`). |
| `get_agent_conversation`      | Get a specific agent conversation by ID.                                                                                                                                                                   |
| `reply_to_agent_conversation` | Send a follow-up message to an existing agent conversation.                                                                                                                                                |

**create\_agent\_conversation parameters**

| Parameter | Type   | Required | Description                                                                                                 |
| --------- | ------ | -------- | ----------------------------------------------------------------------------------------------------------- |
| `feature` | string | Yes      | The platform agent to start a conversation with. One of the feature values listed below.                    |
| `prompt`  | string | Yes      | The initial message to send to the agent.                                                                   |
| `payload` | object | No       | Feature-specific context (for example, the flow or step the agent should work on). See the reference below. |

**create\_agent\_conversation feature and payload reference**

| Feature                | What it does                                                                                                            | Payload fields                                                                                                                        |
| ---------------------- | ----------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------- |
| `assistant`            | General Patchworks assistant for questions and guidance about the platform.                                             | None — pass `{}` or omit.                                                                                                             |
| `implementation-agent` | Builds new integrations end to end, including creating new flows. Use this when you need a flow that doesn't exist yet. | None — pass `{}` or omit.                                                                                                             |
| `connector-builder`    | Builds a new connector, or continues work on an existing one.                                                           | `connector_id` (int, optional) — pass to continue building an existing connector; omit to create a new one.                           |
| `flow-builder`         | Edits a single existing flow.                                                                                           | `flow_id` (int, **required**). The flow builder only edits existing flows — to create a new flow, use `implementation-agent` instead. |
| `map-builder`          | Builds or edits the field mapping for a specific map step in a flow.                                                    | `flow_step_id` (int, **required**), `flow_version_id` (int, optional), `flow_id` (int, optional).                                     |
| `answers-agent`        | Answers questions about your integrations and data.                                                                     | None — pass `{}` or omit.                                                                                                             |
| `documentation-agent`  | Generates documentation, optionally scoped to a specific flow.                                                          | `flow_id` (int, optional) — pass to document a specific flow.                                                                         |
| `health-monitor`       | Investigates the health of a flow or a specific flow run and helps diagnose problems.                                   | `flow_run_id` (string, optional), `flow_id` (int, optional) — pass either to focus the agent on a particular run or flow.             |

Once a conversation is created, use `reply_to_agent_conversation` to continue it with follow-up messages, and `get_agent_conversation` to retrieve its current state.

<br>