Weld
Start free

Harvest Connector

Intuitive time tracking software for teams. Harvest makes it easy to track time to help manage both your projects and workload.

Start for freeView documentation
4.8 on G2
Weld · Data Sources
HarvestLive
clientSynced
projectSynced
userSynced
time_entrySyncing
expenseQueued
Lands insnowflakebigquerydatabricksAgent context over MCP
How it works

Set up in minutes.
No code, no maintenance.

Harvest

Harvest

Data source

Authenticating...
Ready to connect
Overview

Harvest in your warehouse.
Ready for your agents.

Intuitive time tracking software for teams. Harvest makes it easy to track time to help manage both your projects and workload. Weld's Harvest connector is fully managed: it extracts data from Harvest, loads it into your warehouse, and keeps it in sync. The data lands in a clean, documented schema with agent context attached, so analysts can query it with SQL or dbt, and agents can query it over the Weld MCP server without guessing at tables or joins.

AgentsQuery over MCPYour warehouse34 tables, with agent contextWeldExtract, load, keep in syncHarvestSource
Agent context

Harvest data schema.
The map your agents read.

Every table, field and relationship Weld syncs from Harvest is served to agents as context over MCP, so they pick the right tables and join on the right keys.

tables
34
fields
435
relationships
33
Open the schema explorer Open the full schema explorer
Weld MCP server

Ask your agent.
It already knows Harvest.

Connect Claude, Cursor, ChatGPT or your own agent to Weld over MCP. It reads the Harvest context, finds the tables it needs and queries your warehouse.

  • Agent contextHarvest tables, fields and joins
  • QuerySQL against your warehouse
  • OperateCheck, rerun and refresh syncs
Set up the MCP serverClaudeCursorChatGPTGemini
Claude CodeOpus 5
>

How many project rows do we have per client? Use our Harvest data.

Reading the Harvest context first, so the query uses the right tables and keys.

●weld·search_available_references
⎿raw.harvest.project
raw.harvest.client
+ 32 more Harvest tables
●weld·get_reference_schema
⎿project · 27 columns
joins client on client_id
●weld·run_query
⎿select c.id, count(*) from {{raw.harvest.project}} p
join {{raw.harvest.client}} c on p.client_id = c.id
group by 1
submitted
●weld·get_query_status
⎿DONE
●weld·get_query_result
⎿rows returned
✓Answered from Harvest data in your warehouse
>Ask about your Harvest data
Destinations

Load Harvest into any warehouse.
Set up in minutes, no code.

Resources

Everything you need.
To get started.

Documentation

Harvest setup guide

Step-by-step instructions to configure and start syncing.

MCP server

Connect your agents

Set up Claude, Cursor, ChatGPT or Codex against your Weld workspace.

Weld Connect API

Build custom connectors

Use the Weld Connect API to build and publish your own connector.

FAQ

Common questions

Answers to frequently asked questions about Weld connectors.

Weld Connect API

Set it up in code.
One REST API.

Use the Weld Connect REST API to create Harvest connections, configure ELT syncs, and manage pipelines entirely through code. Built for embedding, automation, and CI/CD workflows.

Explore Weld ConnectAPI reference
POST/elt_syncs
▌
FAQ

Common questions.
About connectors.

Something not covered here? A data engineer can walk through your setup with you.

Talk to us

Put your Harvest data to work.

Connect Harvest, pick a warehouse, and your agents can query it today. Free for 14 days, no credit card.

HarvestSource
WeldWeld
MCPAgents
Start for freeTalk to us
Connectors

Explore more connectors.
300+ and counting.

View all connectors