The candor CLI is how a terminal-capable agent works in your financial workspace: stable commands, machine-readable answers, and finance methods it can load when a task calls for them. If your agent can run a shell, it can run your money review.
How it works with your agent
Setup is a conversation, not a wizard. Your agent installs the CLI, runs candor setup, and pauses at each step that belongs to you: approving its access, choosing a plan, connecting accounts. Local credentials use native OS credential storage by default on macOS, Windows, and Linux.
candor open is the agent's home screen: what changed since it last looked, plus what still needs attention. From there it drills down into transactions, balances, recurring charges, budgets, goals, holdings, and debts. Each search returns only the records the question needs, so your agent is never loading your whole financial life at once.
Every answer includes the facts, how fresh they are, anything that may be missing, and the records behind each number. Calculations use those records instead of asking AI to predict a number, and your agent can check the underlying transactions before it repeats a claim to you.
Finance skills teach your agent how the work is done: the disciplined monthly review, duplicate-charge recovery, budget upkeep, trial and renewal watching. Skills load just in time for the task at hand, so the workspace stays lean and the method stays current.
Claude Code, OpenClaw, Hermes, and any other agent that can use a terminal, plus MCP for the ones that speak it. Candor does not care which agent you choose. Your records, budgets, and history stay yours when you switch.
Keep reading
One message to the agent you already use. You see the price before anything private, approve the access and accounts yourself, and your agent is working within minutes.