What it can do

The agent’s abilities come from its tools. Each tool is a typed action with a permission level. This page describes the kinds of tools the agent has and what they let it do.

Query your data

The agent can read your finances through a read-only query tool. This is how it answers questions like “how much did I spend on groceries last month” or “what is my biggest expense this week.”

Two rules make this safe:

• Allowlist. Nothing is queryable unless it has been declared as readable. Models, and the columns within them, are opted in explicitly. Anything not on the list does not exist as far as the tool is concerned.
• Owner-scoped. Every read is filtered to the acting user. The agent cannot read another person’s data, even if it tried to ask for it.

Create and categorise transactions

When writes are enabled, the agent can create and update records: add a transaction, set its category, attach it to a wallet, and link it to the party you paid or were paid by. This is how “add a 12.50 coffee expense to my cash wallet” turns into a real entry. It can also create the wallet, category, or party first if the one you named does not exist yet.

The agent never saves silently. It proposes the change with the details filled in, and the record is created only after you confirm. You can edit any field on that confirmation first, and the saved entry (and its summary) reflects your edits, not the original suggestion.

The same guardrails apply:

• Owner forced to you. On create, the owner is set to the acting user. The agent cannot create data owned by someone else.
• Writable columns only. Only fields that have been marked writable can be set. Sensitive or computed fields are off limits, regardless of what the model asks to write.
• Off by default. Writes are disabled unless the instance turns them on and a user is present. See Permissions and safety.

Move money between wallets

The agent can record a transfer from one of your wallets to another, for example “move 100 from Cash to Bank.” When the two wallets use different currencies it uses the exchange rate you give it, or asks for one if it does not have it. Like any other write, a transfer is proposed first and recorded only after you confirm.

Build reports and visualisations

Beyond a one-line answer, the agent can assemble a richer response: headline figures, tables, and charts, woven with short written sections. Ask for “a report on my spending with graphs” and it queries what it needs and lays out a document in a side canvas. It can chart any data it has pulled, not just a fixed set of breakdowns, so a “pie chart of my top purchases” works the same as “spending by category.” It also has section types for highlighted insights, activity timelines, and progress toward a target.

Import a statement or receipt

When you attach a file, the agent reads it. It can pull the line items out of a bank statement or the details off a receipt and turn them into transactions for you to review and confirm, rather than asking you to type them in.

Ask you when something is unclear

If a request is ambiguous, for example you named a wallet that does not exactly match one of yours, the agent asks instead of guessing, often with a few options to pick from. Your answer continues the task in progress: it remembers what you were doing, so replying “the Main Wallet” finishes the transaction you started rather than being read as a fresh question.

Answer questions and do the arithmetic

The agent can reason over what it reads and explain it back to you. For exact numbers it has a calculator tool, so totals and rates do not depend on the model doing mental math. It also has a clock tool for the current date and relative anchors like “last month,” so date-based questions resolve correctly.

External tools (optional)

The package also defines external tools, such as fetching an allowlisted URL or searching the web through a pluggable driver. These are external-permission tools and are constrained the same way: an allowlist that denies everything by default. A Trakli instance enables them only if it has a reason to.

What it will not do

• It will not read or change another user’s data.
• It will not write anything if writes are disabled on the instance.
• It will not touch fields or models that have not been opted in.
• It will not finalise a create, transfer, or import without your confirmation; every change is proposed first.
• It will not act on a guess about who you are; it acts as the authenticated user, full stop.