mcpField
U
UseHourly

UseHourly

Remote
usehourly.appProductivity0Updated today

About UseHourly

UseHourly is a free, open-source time tracker for freelancers and the AI assistants they use. It combines timers, editable timesheets, invoicing, reporting, and exports, with MCP tools scoped to each user’s data.

Key features

  1. 01Log time and run timers
  2. 02Edit weekly timesheets and calendars
  3. 03Create and track invoices
  4. 04Review yearly hours and earnings
  5. 05Export spreadsheets and account data
  6. 06Import Toggl or Clockify CSVs

Use cases

  1. 01Record work time for a client project
  2. 02Prepare a weekly timesheet for sign-off
  3. 03Create an invoice from billable hours
  4. 04Import existing Toggl or Clockify records

MCP support

UseHourly runs a remote Streamable HTTP MCP server with tools for projects, clients, time entries, timers, timesheets, invoices, and reports. Access uses user tokens with read-only or read-and-write permissions; a local bridge is also documented for stdio-only clients.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://usehourly.app/api/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

All endpoints
Streamable HTTPhttps://usehourly.app/api/mcpAuth required
Packages
npmusehourly-mcp1.1.0

Tools

Details

From the project README

UseHourly

The free, open-source time tracker your AI assistant can use. Connect Claude, ChatGPT, Cursor or any MCP assistant and say "log 3 hours on Acme, then make this week's timesheet for sign-off". Or log your day in seconds yourself: type 1.5, 1:30 or 90m, press Enter, done.

Live: https://usehourly.app · Free to use, no paywall, no ads.

smithery badge

https://github.com/Zontak-LLC/usehourly/raw/main/public/landing/assistant-demo.mp4

The weekly timesheet

What it does

  • Quick entry. Pick a project, type the time the way you would say it, add a note. Or press Start and let the timer count.
  • Weekly timesheet. Every project and every day in one editable grid. Fill several days at once, add more entries to a day, move an entry to another date.
  • Month calendar with per-project totals.
  • Invoices. Turn billable hours into a PDF invoice, track paid and unpaid, re-download any time.
  • Year summary. Hours, earnings, invoiced and received by quarter and month, ready for your accountant.
  • Export. Excel for any week, month or custom date range (for example March 3 to April 20), and one zip with everything: every client, all time.
  • Try it first. "Try it with sample data" on the home page opens a throwaway account filled with a month of work. No sign-up; keep it with one click if you like it.
  • Switching from Toggl or Clockify? Import their CSV export from the Account page. Clients and projects are created for you, and importing the same file twice adds nothing.
  • AI first: works with your assistant (MCP). Any MCP client (Claude, Cursor, VS Code, ChatGPT, your own agent) can log time, start timers, build a client's timesheet, create a private sign-off link and prepare invoices. See below.
  • Free calculator. Hours calculator: add up times and get decimal hours, no account needed.
  • Your data is yours. Per-user isolation; everything you log can be exported at any time.

Use it from any AI assistant (MCP server)

UseHourly runs a remote Model Context Protocol server at https://usehourly.app/api/mcp (Streamable HTTP). It has 32 tools for projects, clients, time entries, timers, timesheets, invoices and reports; each one runs as you, so it only sees your own data. Create a token on Account > AI assistants (read-only or read and write, with an expiry), then:

claude mcp add --transport http usehourly https://usehourly.app/api/mcp --header "Authorization: Bearer <your token>"

That example is for Claude Code. Other clients just need the URL and an Authorization: Bearer <token> header in their MCP settings.

Cursor (~/.cursor/mcp.json):

{ "mcpServers": { "usehourly": { "url": "https://usehourly.app/api/mcp", "headers": { "Authorization": "Bearer <your token>" } } } }

VS Code (.vscode/mcp.json):

{ "servers": { "usehourly": { "type": "http", "url": "https://usehourly.app/api/mcp", "headers": { "Authorization": "Bearer <your token>" } } } }

For clients that can only run local (stdio) servers, such as Claude Desktop, there is a small bridge published on npm (usehourly-mcp): set the command to npx with args ["-y", "usehourly-mcp"] and USEHOURLY_API_TOKEN to your token.

Then ask: "Log 2 hours on the Acme website for yesterday" or "Make an invoice for Acme for September."

Stack

Next.js 16 (App Router) · React 19 · Tailwind CSS v4 · shadcn/ui · NextAuth v5 (credentials) · tRPC v11 · Prisma 7 with Cloudflare D1 · Vitest · Biome · deployed to Cloudflare Workers with OpenNext.

Run it locally

You need Node 22+.

npm install                 # also runs `prisma generate`

cp wrangler.example.jsonc wrangler.jsonc   # then set account_id and database_id (see Deploy)

# .env.local and .dev.vars (keep the two in sync):
#   AUTH_SECRET=<a long random string, e.g. `openssl rand -base64 32`>
#   AUTH_URL=http://localhost:3028

npm run d1:migrate:local    # create the local D1 database
npm run dev                 # http://localhost:3028

Register an account on the sign-up page and you are in.

npm test                    # scenario tests (Vitest)
npx biome check             # lint and format
npx tsc --noEmit            # type check

Deploy

It runs on Cloudflare Workers with a D1 database:

  1. cp wrangler.example.jsonc wrangler.jsonc and fill in your Cloudflare account id and a D1 database (npx wrangler d1 create usehourly-db). wrangler.jsonc is git-ignored, so your ids stay out of the repo.
  2. Put your public address in .env.production.local: NEXT_PUBLIC_APP_URL=https://app.example.com (it is baked into the build) and set the same value as AUTH_URL in wrangler.jsonc.
  3. npm run d1:migrate:remote
  4. npx wrangler secret put AUTH_SECRET
  5. npm run deploy
Email (optional)

Password reset and email confirmation need an email service. UseHourly sends through Resend: verify your sending domain there, then set two values on the Worker:

npx wrangler secret put RESEND_API_KEY
npx wrangler secret put EMAIL_FROM        # for example: UseHourly <[email protected]>

Without them the app works as before, and the forgot-password page says email isn't set up.

Deploy from CircleCI

.circleci/config.yml runs the checks on every push and, on main, applies the D1 migrations and deploys. The deploy token is a secret, so keep it out of the project settings: create a context named usehourly-production (Organization Settings > Contexts), add an expression restriction pipeline.git.branch == "main", and put the token in it. Only the deploy job on main can read it.

Context variableValue
CLOUDFLARE_API_TOKENa token with Workers Scripts: Edit and D1: Edit

In the CircleCI project settings add these environment variables:

VariableValue
CLOUDFLARE_ACCOUNT_IDyour Cloudflare account id
D1_DATABASE_IDthe id of your D1 database
APP_URLyour public address, for example https://app.example.com

Optional: WORKER_NAME, D1_DATABASE_NAME, CUSTOM_DOMAIN. AUTH_SECRET stays a Worker secret (npx wrangler secret put AUTH_SECRET, once); CI never sees it. scripts/ci-wrangler-config.mjs writes wrangler.jsonc from these variables, so no account ids live in the repository.

Support

UseHourly is free. If it saves you time you can sponsor it on GitHub.

Supporters

Thank you to everyone who helps keep UseHourly free:

Be the first: sponsor UseHourly on GitHub.

Credits and license

Built on the excellent next-shadcn-admin-dashboard template by Mohammed Arham Khan.

Released under the MIT License. The code is open; the name "UseHourly" and its logo are not part of the license.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.