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

With no annotations, the description carries the full behavioral burden, and it does so thoroughly. It discloses that the tool does NOT check that the sheet balances and why (retained earnings is the balancing figure), warns against claiming it was checked for consistency, states it is free and instant, and describes the return values (derived sheet, rejected figures, completeness status). It also clarifies unit conventions (major units) and the dormant filing warning. This is comprehensive behavioral disclosure far beyond minimal assumptions.

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 description is relatively long but every sentence carries substantive information—purpose, returns, limitations, usage, units, and parameter guidance are all present with no filler. It is front-loaded with the purpose and immediately clarifies a critical limitation (no balance check). While not ultra-terse, it is efficiently organized and each sentence earns its place, so a 4 is warranted rather than a 5 for extreme economy.

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?

Given the tool's complexity (3 parameters, nested figures object, 16 required sub-fields), the description provides a complete picture: what it returns, its limitations, when to call it, how to handle units, and important parameter relationships. With no output schema, it adequately describes the output components. The description suffices for an agent to correctly invoke the tool and interpret results, leaving no critical gaps.

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?

Despite 100% schema description coverage, the description adds valuable semantics beyond the schema: it explains that directorsLoan is part of creditors (never an addition), often the largest line, and repayable tax-free; it notes that lastFiledAs must come directly from lookup_company and not be guessed; and it reiterates major units globally. These enrich the bare schema definitions, justifying a score above the baseline of 3. It doesn't exhaustively detail every parameter but covers the most decision-relevant nuances.

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?

The description opens with a specific verb and resource: 'Derive a small-company balance sheet (micro-entity, in UK terms) from raw figures and validate it.' This clearly identifies the tool's function and differentiates it from siblings like explain_accounts (which presumably interprets) and totals_to_figures (which likely prepares figures). It also states precise scope and outputs, leaving no ambiguity about the tool's purpose.

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?

The description explicitly advises 'call it after every change to the figures' and gives the condition for the dormant warning ('when figures showing activity are about to be filed dormant again'), which guides when to invoke it. It also instructs to pass lastFiledAs from lookup_company. However, it does not explicitly name alternative tools or state when NOT to use this one, only implying it is a safe always-call tool. The guidance is clear but lacks explicit exclusion criteria, so a 4 is appropriate.

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