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

With no annotations, the description carries the full burden and delivers: the tab persists across conversation turns, 'encerrar' only on request, and 'consequencia' behavior including that omission triggers a user prompt. It exposes non-obvious mechanics (ref doesn't appear in the response, roteiro halts on first failed 'conferir'). This is exactly the extra context annotations would otherwise provide.

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?

Front-loads purpose and the 'computador' distinction before drilling into actions. Density is high but nearly every sentence earns its place for a nine-action tool. Slightly heavy overall, which keeps it just shy of perfect.

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 complex multi-action tool with nested objects and no output schema, the description supplies the missing pieces: that 'navegar' returns the page tree with refs (return shape), the roteiro result contract, and the persistence model. An agent has everything needed to invoke it correctly.

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 only 25%, so the description must compensate and does: url, ref semantics (must come from a prior reading, never invented, aim-only), ref-less 'digitar' writing to focus, 'enviar' pressing Enter, 'nome' after navegar in a script, and the full 'passos' structure. This is far beyond what the sparse schema documents.

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?

Opens with a specific verb+resource ('Abre e opera uma aba própria no Chrome') and enumerates the operations (lê, clica, preenche, rola). It explicitly distinguishes itself from the sibling 'computador' by routing desktop-program tasks there, so an agent can tell the tools apart without opening schemas.

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 per-action when-to-use guidance ('Comece com navegar', 'snapshot lê a página de novo', 'imagens... só quando o texto não bastar'), plus when to escalate to 'roteiro'. It even states exclusion rules (use 'computador' for desktop programs) and the retry heuristic ('se nada mudou, procure o bloqueio em vez de repetir').

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?

With no annotations present, the description carries the full behavioral burden and does so richly. It discloses post-gesture verdicts (`confirmado`, `mudou`, `sem_efeito_aparente`, `nao_da_para_confirmar`), the `na_janela` placement check, approval requirements for the first gesture each turn, refusal of all actions after `pedir_a_pessoa` until the human returns, and special behaviors like apps opening behind or programs opening in the person's window. It also explains that `fechar` behaves like the X and that programs may still ask to save.

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 long but front-loaded: it opens with what the tool does and when to use it before moving into workflow details. Each paragraph has a clear theme (orientation, verdicts, programs, text, games, approval), and most sentences add distinct operational value. It is slightly dense and could use tighter grouping, but the length is defensible for a 30-parameter computer-control tool.

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 30 parameters, no annotations, no output schema, and high schema coverage, the description supplies the missing behavioral and workflow context. It covers approval, human handoff with `pedir_a_pessoa`, communication via `dizer`, when to call `encerrar`, and special cases like games, carousels, and animated menus. Nothing essential for correct invocation appears to be 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 description coverage is 93%, so the baseline is 3, but the description adds practical usage semantics beyond the schema. It explains how to orient with `ver` and `marcar: true`, to target by `elemento: n` or by screenshot coordinates rather than guessing, and to use `ritmo` differently for websites versus local apps. It also notes browser-address-bar behavior (press `delete` before `enter`) and batching gestures in one turn with failure halting subsequent gestures, which enriches several parameters beyond their schema descriptions.

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 first sentence gives a concrete verb and resource: it sees the screen and uses real mouse and keyboard. It immediately distinguishes this tool from the sibling `browser`, saying site tasks are better handled there and this tool covers what the browser cannot reach. An agent can tell it apart from `browser` and `gravacao` without opening any 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?

The description explicitly states when to use this tool versus the alternative: for computer programs and anything the browser does not reach, while website tasks should go through `browser`. It also gives detailed routing guidance for sub-actions like `ver` with `marcar: true`, `elementos`, `ler` with `procurar`, `focar` versus `abrir`, and `pedir_a_pessoa` for login or phone-code steps. When-not conditions are covered through the handoff to the human and the browser alternative.

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 declare readOnlyHint=false. The description adds real behavioral context beyond them: that `iniciar` requires a person's request and specific approval, the privacy caution against publishing/sharing without reviewing personal data, and the efficiency constraint of not pulling hundreds of frames. It stops short of describing the return format of `estado` or the reversibility of edits.

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?

Front-loaded with the overall purpose, then organized by action using backticked names, followed by workflow and privacy notes. Dense but each sentence carries information, though several clauses re-state schema-level constraints.

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 13-parameter, five-action tool with no output schema, it covers each action, the approval prerequisite, the `ver` workflow, and a privacy caution. `estado` is only described as 'consulta' with no indication of what it returns, which is the main remaining gap without an output schema to rely on.

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 92%, so the schema already documents essentially every parameter (de, ate, area, modo, trechos, etc.). The description only reinforces how de/ate and area combine for region investigation and restates the modo 'mudancas' 30 s limit, which is already in the schema. Baseline 3 is correct when the schema does the heavy lifting.

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 clear purpose ('Grava, inspeciona e edita uma captura local da tela') and enumerates all five sub-actions with a specific meaning for each (iniciar/parar/estado/ver/montar). An agent can tell what the tool does and how each action differs. It does not, however, distinguish this tool from the sibling tools 'browser' and 'computador'.

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 concrete when-to-use guidance: 'iniciar' requires a person's request and specific approval, `ver` should be used with de/ate and area to investigate a region and refine the interval across calls rather than requesting hundreds of images, and 'mudancas' mode suits something that appears and disappears. No explicit exclusion against the sibling tools, so it falls short of a 5.

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