mcpField
C
Candor Finance

Candor Finance

Remote
candor.moneyFinance2Updated 5d ago

About Candor Finance

Candor Finance brings personal financial accounts, goals, budgets, and weekly reviews to an AI agent a person already uses, while retaining plans and decisions. It calculates figures from account records, flags potential fees or missed refunds, and never moves money.

Key features

  1. 01Connect bank, card, loan, and investment accounts
  2. 02Set financial goals and budgets
  3. 03Review spending, bills, debt, and savings weekly
  4. 04Find fees, price increases, and possible missed refunds
  5. 05Record agent actions and their reasons

Use cases

  1. 01Create savings goals and budgets from account data
  2. 02Review spending, bills, and investments each week
  3. 03Look for fees, price increases, or missed refunds
  4. 04Continue a financial plan across different agents

MCP support

Candor's hosted MCP server uses OAuth to provide agents with computed account figures, goals, budgets, rules, notes, and weekly reviews, and lets them read and save workspace data. It cannot move money or change financial institution accounts.

Configuration

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

Endpointhttps://api.candor.money/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.

Tools

Details

From the project README

Candor Finance

Candor is a personal finance app that helps you reach your money goals with ChatGPT, Claude, or the AI agent you already use. Your agent works from your real bank, card, loan, and investment accounts: it sets up goals and a budget with you, finds money you are losing to fees, price increases, or missed refunds, and remembers your plan and decisions between conversations. Every week Candor reviews your spending, bills, debt, savings, and investments and tells your agent what's worth talking through and why. Candor never moves money.

Reach your money goals with your AI agent

Start free trial · See how it works · Security

  • Works with ChatGPT, Codex, Claude, Claude Code, Grok Bot, Gemini CLI, OpenClaw, Hermes, and any agent that supports MCP and Agent Skills.
  • US accounts connect read-only on Candor's website.
  • Your data is not sold or used to train models.
  • Every agent action is recorded with its reason.

Add Candor to your agent

AgentWhere to get it
ChatGPT and CodexCandor Finance in ChatGPT's directory
ClaudeCandor Finance in Claude's directory
Claude CodeInstall from this repo's marketplace
Gemini CLI, OpenClaw, Hermes, and othersSend your agent START.md

Your agent asks you to approve its access once, then takes its first look. Setup guides for each agent are at candor.money/connect.

What stays true

  • Candor never moves money. Accounts connect read-only through a bank-grade connection provider. Your agent acts outside Candor only with your okay.
  • Figures come from your records. Candor computes totals, budgets, and goal paces from your accounts, so your agent isn't guessing a plausible number.
  • You can see what your agent did and why. Every action is recorded with its reason, and your plans and decisions stay saved when you switch agents.
  • Your data isn't sold or used to train models. No ads. You're the customer. See Privacy.

What this plugin runs and sends

  • No hooks, commands, agents, or background processes. The plugin adds skills and one MCP server, nothing else.
  • Skills are instructions. Each skill is Markdown. The full checkup skill also includes scripts/review-sweep.mjs, a helper the agent can run with Node on Candor responses it saved. It reads only the files it is given, prints a summary, and makes no network requests.
  • One MCP server, Candor's own. It is the hosted server at https://api.candor.money/mcp, which your agent authorizes with OAuth. Through it the agent reads and saves your Candor workspace. It cannot move money or change anything at a financial institution.
  • Two non-secret headers. Requests to that server carry Candor-Agent-Package-Version and, where the host supports it, Candor-Agent-Package-Route, naming the package version and where it was installed from. Neither holds a credential or personal data.
  • No local secrets. The plugin reads no environment variables, credentials, or local files beyond those the agent hands the helper script.
  • openclaw/ is separate. That subtree is a CLI-based skill for OpenClaw only and is not part of this plugin.

For agents: connection materials

The official package for Candor. It gives the AI agent you already use the person's records, the figures Candor computes from them, the goals, budget, rules, and notes it keeps, and Candor's weekly review, plus one skill per job: getting started, planning a goal, budgeting, weekly check-ins, a full checkup, sorting transactions, finding money, tax prep, and adding records.

Give your agent START.md. It has the MCP connection, skill download, and supported package options. The agent uses its own harness's setup workflow and asks you only for steps it cannot complete.

  • MCP: https://api.candor.money/mcp, Streamable HTTP with OAuth discovery and dynamic client registration. Your harness obtains its own client ID; no shared client secret is needed.
  • Skills: every directory under skills/. candor-finance is the front door and the rest are workflows. START also links a single-skill ZIP that carries all of them for hosts that upload one skill.
  • Supported packages: Agent Plugins 1.0.0 (plugin.json, skills/, mcp.json), Claude and Codex marketplaces (.claude-plugin/ and .codex-plugin/), and a Gemini CLI extension (gemini-extension.json). Use whichever format your harness supports, or install skill and MCP separately.

Your harness owns installation, authorization, skill locations, and reloads. Keep working installations. Once the skill and tools are available, call candor_open for MCP or run candor open for the CLI, then continue into the user's financial task. Use a private runtime or per-user account isolation before authorizing personal financial access on a shared host.

Grok Bot installs the cursor/ plugin from the Cursor Marketplace. It pairs the same MCP-based skill with the remote MCP server and is listed for Grok Bot only; .cursor-plugin/marketplace.json points the marketplace at it.

The root skill uses MCP, including for capable OpenClaw and Hermes hosts. The separate openclaw/skills/ projection uses the Candor CLI and is only for CLI installations. Keep the skill matched to the connection it operates. Candor continues to enforce sign-in, access, and financial-data permissions.

Manage your Candor account and subscription at app.candor.money. When Candor gives the agent a secure account link, it explains the required step and opens that exact page.

Except for the scoped MIT-0 OpenClaw skill subtree and the scoped MIT cursor/ plugin, Copyright 2026 Candor Systems, Inc. All rights reserved. See LICENSE.

Release notes

See the changelog for changes by package version.

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.