# Tools

The MCP server exposes tools organised into three groups. Groups are loaded or omitted as a whole — you cannot enable or disable individual tools within a group. All three groups are enabled by default; use the `LLMBASE_MCP_TOOLS` environment variable to restrict which groups are loaded.



## Tool groups

| Friendly name        | Group key     | Tools                                                                                                                                         | Data source                        |
|----------------------|---------------|-----------------------------------------------------------------------------------------------------------------------------------------------|------------------------------------|
| Documentation search | `collections` | `search_python-docs`, `search_python-examples`, `search_cloud_coverage`, `search_sql-docs`                                                    | Vector-indexed docs (pgvector)     |
| Query Based          | `queries`     | `search_symbol_universe`                                                                                                                      | Live OneTick WebAPI query          |
| Reference data       | `csv`         | `list_databases`, `db_tables`, `db_table_schemas`, `db_field_descriptions`,`db_field_enumerations`, `calendar_holidays`, `trading_hours` | Flat text files in `CSV_DATA_PATH` |

See `Documentation Search Tools`, `Symbol Universe Tool`, and `Reference Data Tools` for a description of each individual tool and when to use it.



## Enabling and disabling tool groups

Set `LLMBASE_MCP_TOOLS` to a comma-separated subset to disable groups you don’t need.

For example, if you do not have OneTick WebAPI credentials, omit the symbol universe group to
get started without it:

```
LLMBASE_MCP_TOOLS=collections,csv
```

You can add it later once credentials are available. See `OneTick WebAPI authentication` for
the full auth configuration.

## Bringing your own data

The pre-configured tools reflect an OneTick Cloud setup. Both the documentation search index and the reference data can be replaced or extended for other deployments:

* **Documentation search** — register additional collections (document sets) that are indexed and exposed as new `search_*` tools alongside the built-in ones. See `Custom Search Tools`.
* **Reference data** — replace the bundled CSV files by pointing `CSV_DATA_PATH` at your own directory. See `Reference Data Tools`.