Zovo Expense Report

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare readOnlyHint=true and destructiveHint=false, and the description adds substantial behavior beyond that: it names the returned artifacts (PDF with employee, department, period, line table, category subtotals, advances, total due, signature block; Excel file), the policy flags raised (missing receipts, duplicates, weekends/holidays, out-of-period, alcohol in meals, per-line cap), and the handling rules for duplicates, totals mismatches, and unconverted currencies. No behavior contradicts the read-only annotation, since the tool produces documents rather than mutating external systems.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Is the description appropriately sized, front-loaded, and free of redundancy?

It is front-loaded with the use case and contains little pure filler, but it is one dense run-on paragraph chained with semicolons and commas, including a nested bracket list that is hard to parse on first read. The formatting rules, ambiguity handling, and output description would be far more scannable as short labeled sections.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 17-parameter tool with one required parameter, the description covers the required 'lines' input in depth, discloses the return set (even though an output schema exists), and states limits and side effects precisely. Optional parameters are fully documented in the schema, so nothing an agent needs to invoke this correctly is missing.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the baseline is 3; the description earns extra by prescribing the preferred object shape per expense, giving the multi-night-stay alternatives ({merchant, amount, quantity} vs {merchant, quantity, unit_price}), and defining free text as a strict fallback with a named failure bucket ('Not counted, check these'). It leaves most optional fields (advance, approver, timezone, date_order, receipt_required_over) to the schema, which is acceptable given full coverage.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Does the description clearly state what the tool does and how it differs from similar tools?

States a concrete verb and artifact ('generate an expense report... from their receipts'), enumerates the accepted line shapes, and closes with explicit non-capabilities ('does not read images itself, look up exchange rates, or submit anything'). The repeated emphasis on sending structured lines yourself implicitly separates it from the sibling generate_expense_report_from_file, which takes a file instead of caller-built lines.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Does the description explain when to use this tool, when not to, or what alternatives exist?

Opens with an explicit trigger ('Use this when the user wants an expense report or reimbursement form'), adds recognized user phrases, and gives a real exclusion: 'A bank or card statement is not an expense report: ask the user which charges are business expenses and send only those.' It also states what to do when input is ambiguous rather than guessing.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare readOnly, non-destructive and closed-world, but the description adds substantial behavior beyond them: the refusal path for bank statements with the recommended recovery (ask which charges are business expenses), the list of recognized expense-tool export formats, the returned artifacts (PDF, Excel, policy flags), and the strict-line parsing behavior where ambiguous lines are handed back for resubmission.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Is the description appropriately sized, front-loaded, and free of redundancy?

The usage condition and the sibling alternative are front-loaded in the first sentence, and the refusal/limitation clauses are grouped afterward. It is dense and long, and the sentence about pasted text sent to generate_expense_report is somewhat tangential to this tool, but nearly every clause carries actionable information.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 17-parameter tool with a nested file object, the description covers input format acceptance, refusal cases, recovery instructions, and output artifacts, with the output schema handling return details. An agent has everything needed to decide whether to call it and what to expect.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100% and each of the 17 parameters carries its own explanation and defaults, so the schema does the heavy lifting. The description contributes context about the recognized input columns (Date, Merchant, Category, Amount, Miles, etc.) but says nothing about the optional report-shaping parameters, which is acceptable at baseline 3.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource ('generate an expense report ... from an attached CSV or Excel list') and explicitly names the sibling generate_expense_report as the alternative path. An agent can distinguish the two tools without opening either schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Does the description explain when to use this tool, when not to, or what alternatives exist?

Gives explicit positive conditions (user attaches a CSV/.xlsx expense list), an explicit alternative ('if you can read the file yourself, prefer sending its expense lines as objects to generate_expense_report'), and explicit exclusions (bank/card statement exports are refused; receipt images and PDFs are not read; nothing is submitted to an employer system). The routing decision is fully determined by the text.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.