Server Quality Checklist

Tool Scores

• cancel_sessionB
  3.1/5.0
  Behavior2/5

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

  No annotations are provided, so the description carries the full burden of behavioral disclosure. It mentions cancellation and the return shape, but does not explain side effects (e.g., whether in-flight requests are aborted, whether the session is permanently deleted or recoverable), authentication requirements, or idempotency. For a mutating action without annotation coverage, this is a significant gap.

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

  Conciseness5/5

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

  The description is a single, concise sentence that front-loads the action and immediately states the return value. There is no wasted text, and it is appropriately sized for a simple cancellation tool.

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

  Completeness3/5

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

  Given the low complexity (one parameter, no output schema, no annotations), the description provides the essential action and return format. It is minimally sufficient for an agent to invoke the tool, but lacks contextual details like side effects or error handling. For a mutation tool, this is adequate but not rich; a 3 reflects the missing behavioral context without over-penalizing given the tool's simplicity.

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

  Parameters3/5

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

  Schema description coverage is 100%, so the parameter 'session_id' is fully documented in the schema. The description adds no extra meaning about the parameter (e.g., format, constraints, or how to obtain it). Per the rubric, baseline 3 applies when schema covers the parameter, and the description does not need to repeat it.

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

  Purpose4/5

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

  The description states a specific verb '取消' (cancel) and a clear resource '标注会话' (annotation session). This distinguishes it from siblings like 'annotate_document' and 'wait_for_annotations', though it does not explicitly name an alternative like the highest-quality examples. The purpose is unambiguous.

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

  Usage Guidelines2/5

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

  The description gives no guidance on when to use this tool versus alternatives. It only states the action and return value. It implies usage for canceling a session, but provides no exclusion criteria, prerequisites, or references to related tools like 'wait_for_annotations' or 'preview_edits'.

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

• read_locationA
  4/5.0
  Behavior4/5

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

  With no annotations, the description carries the full burden of behavioral disclosure. The verb '读取' clearly implies a read-only operation, and the explicit return shape {loc, text, context} tells the agent what to expect. It could additionally mention failure behavior or permissions, but the non-destructive nature is evident.

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

  Conciseness5/5

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

  The description is a single, front-loaded sentence that communicates purpose, usage context, and return shape without any wasted words. Every part contributes to helping an agent decide whether and how to call the tool.

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

  Completeness4/5

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

  For a read-only tool with two well-documented parameters, the description is mostly complete. It compensates for the lack of an output schema by specifying the return shape. It does not elaborate on the six loc kinds, but the schema enum already covers that, so nothing critical is missing.

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

  Parameters3/5

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

  Schema coverage is 100%, so both parameters are already documented in the schema. The description adds no additional semantic information about the parameters themselves; it only describes the return value. The schema's descriptions of path and loc are sufficient, so a baseline score of 3 is appropriate.

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

  Purpose4/5

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

  The description clearly states the verb (read), the resource (a specified location), and the scope (original text plus adjacent context), making the tool's purpose understandable. It does not explicitly differentiate from sibling tools like read_structure, but the focus on content and context is distinct enough.

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

  Usage Guidelines4/5

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

  The description gives a clear usage context: it should be used to check the original text and surrounding context before generating new content. It does not mention alternatives or when not to use this tool, but the stated scenario provides actionable guidance.

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

• read_structureA
  4/5.0
  Behavior4/5

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

  With no annotations provided, the description carries the burden of behavioral disclosure. It clearly communicates that this is a read operation and specifies the exact return shape per format: paragraph-level {loc,text} for docx, sheet dimensions/sampling for xlsx, and shape text for pptx. It stops short of describing error behavior or limits beyond the schema.

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

  Conciseness5/5

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

  The description is compact and information-dense, covering three file formats and the return envelope in a single sentence. Every clause earns its place with no filler or repetition.

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

  Completeness4/5

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

  There is no output schema, but the description compensates by explicitly stating the return structure '{format, items|sheets|slides}' and per-format content. It is complete enough for an agent to know what to expect. Minor gaps like max_items interaction with sampling are not critical for basic invocation.

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

  Parameters3/5

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

  Schema description coverage is 100%: both 'path' and 'max_items' are described in the schema. The tool description adds no parameter-level detail beyond what the schema already provides, so the baseline score of 3 is appropriate.

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

  Purpose5/5

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

  The description states a specific verb ('read') and resource ('document outline') and goes beyond that by detailing format-specific behavior for docx, xlsx, and pptx. This clearly distinguishes it from read_location and other sibling tools.

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

  Usage Guidelines3/5

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

  The description implies this tool is for inspecting document structure before performing edits or annotations, but it does not explicitly name alternatives or state when not to use it. The format breakdown gives implicit context, but there is no direct when/when-not guidance.

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

• template_replaceA
  4.2/5.0
  Behavior4/5

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

  描述揭示了重要的非显而易见行为：未提供值的变量保持原样并在 missing_vars 中列出；同段落多处占位符按从右到左应用保证坐标正确；跨 run 断裂的占位符也能正确替换。这些细节超出 schema 和 annotations 提供的范围（annotations 未提供，schema 未提及这些行为）。

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

  Conciseness5/5

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

  描述紧凑且信息密集，每句话都有明确用途。核心行为在前两个分句中，边界情况（跨 run、从右到左）透明披露，返回结构也说明了。没有冗余。

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

  Completeness4/5

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

  描述涵盖了返回结构（matched, replaced, missing_vars, applied）、边界情况行为和参数含义。但没有明确说明失败模式（如文件不存在、权限问题）或是否修改原文件或生成新文件。考虑到描述已经提供大量行为细节，缺少这些不影响基本使用。

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

  Parameters3/5

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

  Schema 描述覆盖率 100%，所有参数在 schema 中都有描述。描述本身没有为参数添加额外语义，但 variables 参数的行为（未提供值则保持原样）在描述中提到了，这补充了 schema 描述。整体上 schema 已承担主要负担，描述添加了边际价值。

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

  Purpose5/5

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

  描述明确说明工具行为：批量替换 docx 文档中的 {{变量}} 模板占位符，并提到返回值。与兄弟工具如 find_replace 和 apply_edit 有区分（模板占位符替换 vs 通用查找替换/编辑）。

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

  Usage Guidelines4/5

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

  描述隐含了使用场景（模板填充、缺失变量报告），但没有明确说明何时使用 vs 兄弟工具（如 find_replace）。没有提供排除条件，但通过描述变量映射和缺失变量行为暗示了用途。

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

• preview_editsA
  4.3/5.0
  Behavior4/5

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

  With no annotations provided, the description carries full behavioral disclosure and does a good job: it explicitly states the operation is in-memory only and does not write to disk, and that edits are applied sequentially to the same document. It stops short of describing failure modes or precise before/after semantics for edge cases such as delete.

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

  Conciseness4/5

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

  The main safety and purpose information is front-loaded in a compact opening phrase, followed by use case, parameter summary, and return shape. The return format is stated twice, which is mildly redundant, but overall the description is appropriately sized and scannable.

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

  Completeness4/5

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

  For a preview tool with no output schema, the description covers what it returns, how it applies edits, and the critical no-write behavior, and it points to apply_edit as the actual mutation path. Minor gaps remain around edge-case behavior such as delete-mode results or invalid locations, but nothing essential blocks correct selection and invocation.

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

  Parameters4/5

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

  The schema already documents all parameters with 100% coverage, so the baseline is 3. The description adds meaningful semantics beyond the schema by clarifying that the edits list is applied sequentially in memory to the same document, and by summarizing the element shape as {loc, new_content?, mode?, style?}.

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

  Purpose5/5

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

  The description uses a specific verb and resource: batch preview edits, memory-only, with no write to disk. It explicitly returns per-edit {loc, before, after} and differentiates itself from the sibling apply_edit by positioning itself as a pre-check step.

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

  Usage Guidelines4/5

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

  It clearly states when to use this tool: before actually applying edits via apply_edit, to verify whether a batch of modifications is as expected. It names apply_edit as the real-write alternative, though it does not enumerate when-not-to-use conditions or other sibling tools.

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

• find_replaceA
  4.4/5.0
  Behavior4/5

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

  With no annotations, the description carries the behavioral burden and does well: it discloses plain-string matching, table-cell coverage, case-insensitive behavior when match_case=false, verbatim insertion of replacement text, and the return shape. It does not explicitly warn that the file is modified in place or discuss reversibility, but the core mutation semantics are clear.

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

  Conciseness5/5

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

  A single dense sentence with no filler. Every clause adds information: scope, regex exclusion, table-cell inclusion, case sensitivity, replacement semantics, and return contract. The most important operation and scope come first.

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

  Completeness4/5

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

  The description covers the action, scope, case behavior, replacement semantics, and return values, which is enough for an agent to invoke it correctly. It lacks only an explicit statement about whether the document is overwritten in place or how to recover, but the core usage context is complete.

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

  Parameters4/5

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

  Schema coverage is 100%, so the baseline is 3, but the description adds value beyond the schema: it clarifies that match_case=false means case-insensitive, that replacement text is inserted as-is, and that the result includes matched/replaced counts and locations. This deepens understanding of the parameters' runtime behavior.

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

  Purpose5/5

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

  The description names a concrete operation: full-text find-and-replace in .docx files. It explicitly scopes the behavior as plain-string (non-regex) and includes table cells, which clearly distinguishes it from related tools like template_replace or scoped edits.

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

  Usage Guidelines4/5

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

  The description gives clear context: it is for whole-document plain-string replacement with an optional case-sensitivity flag. It does not explicitly name alternatives or state when not to use it, but the intended use case is unambiguous.

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

• list_versionsA
  4.5/5.0
  Behavior5/5

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

  With no annotations provided, the description carries the full behavioral burden and does so thoroughly: it reveals the snapshot-chain model, version merging per annotation round, ordering (new versions first), empty-list behavior, and the meaning of each desc as 'state before that annotation round'. This goes far beyond what the schema alone conveys.

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

  Conciseness5/5

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

  The description is dense but every clause earns its place: purpose, snapshot semantics, output shape, desc meaning, and companion tool. It is front-loaded with the core action and maintains a logical flow from behavior to return format to usage.

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

  Completeness5/5

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

  For a tool with one required parameter, no output schema, and no annotations, this description is complete: it explains the return fields, empty result behavior, version semantics, and how to combine with restore_version. An agent can correctly invoke it and interpret results without additional context.

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

  Parameters3/5

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

  The input schema already has 100% coverage for the single 'path' parameter, so the baseline is 3. The description does not add parameter-specific semantics beyond the schema, but none are strictly needed here.

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

  Purpose5/5

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

  The description starts with a specific verb and resource: '列出文档的版本历史' (list document version history). It further clarifies semantics with the snapshot-chain explanation, output array shape, and explicitly names restore_version as the companion tool, so an agent can distinguish it from siblings like restore_version or read_location.

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

  Usage Guidelines4/5

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

  The description clearly indicates this tool is for reading version history and works together with restore_version for arbitrary rollback, giving a concrete workflow context. However, it does not explicitly state when not to use it, e.g., for reading current document content instead of history.

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

• restore_versionA
  4.5/5.0
  Behavior5/5

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

  With no annotations provided, the description carries full responsibility and does so thoroughly: it discloses that no snapshot is taken before rollback, version history will not increase, later modifications are lost, the rollback cannot be undone via a new snapshot, and old annotation loc indices may become invalid.

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

  Conciseness5/5

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

  The description is dense but well-structured: the core action is bolded and front-loaded, followed by the critical warning, the post-rollback guidance, and the return shape. Every sentence earns its place, with no redundant filler.

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

  Completeness5/5

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

  For a destructive version-restore tool with no annotations and no output schema, the description is remarkably complete: it covers the action, the exact behavior regarding version history, data loss, irreversibility, follow-up steps, and the return payload.

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

  Parameters3/5

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

  Schema description coverage is 100%, so the baseline is 3; the schema already explains path, index, and session_id. The description adds behavioral context around the index parameter (it comes from list_versions) but does not materially extend the schema's parameter documentation.

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

  Purpose5/5

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

  The description states a specific verb ('直接回退' / directly restore), a clear resource (document version from list_versions), and the exact result ('document is overwritten with that version's content'). It also names sibling tools list_versions and read_structure, making the tool's role distinct from listing or editing.

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

  Usage Guidelines4/5

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

  The description clearly identifies list_versions as the source for the version index and instructs the agent to re-run read_structure after rollback before continuing edits. It warns about data loss and irreversibility, but does not explicitly contrast this tool with alternatives like apply_edit or template_replace for non-restore modifications.

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

• wait_for_annotationsA
  4.5/5.0
  Behavior4/5

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

  With no annotations provided, the description carries the full behavioral burden and delivers: blocking semantics, server-side session persistence across calls, no cancellation on truncation, status expansion, and automatic window reload behavior. However, it contradicts the input-schema default for timeout_seconds (description says 'never time out' by default; schema says 'default 1800'). This inconsistency creates agent-facing ambiguity about the actual default wait behavior.

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

  Conciseness4/5

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

  The paragraph is long and dense but nearly every clause carries a distinct behavioral rule spanning multiple statuses and call-pattern scenarios, so the length is justified. Core blocking behavior and default timeout are front-loaded, but the wall of text could be better structured into bullets or status-to-action pairs for faster parsing.

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

  Completeness4/5

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

  For a 2-param tool with no output schema and no annotations, the description covers return shape, per-status meaning, follow-up actions, and session persistence — essentially everything needed for correct invocation. The only substantive gap is the timeout-default contradiction with the schema, which leaves an agent uncertain about the real server-side default.

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

  Parameters4/5

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

  Schema description coverage is 100%, so the baseline is 3, but the description adds meaningful semantics beyond the schema: when and why to pass timeout_seconds (client call limits), what to do if the call is truncated (call the tool again), and that session_id must be forwarded into apply_edit/template_replace/find_replace/restore_version. The added value is slightly offset by the timeout-default contradiction with the schema.

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

  Purpose5/5

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

  States a specific verb and resource: blocks waiting for an annotation session to complete and returns the user's submitted annotations. Explicitly lists the return shape and all possible status values (done | cancelled | closed | timeout), which clearly distinguishes it from sibling edit/apply tools like apply_edit or find_replace. No tautology and no ambiguity.

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

  Usage Guidelines5/5

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

  Provides exhaustive when-to-use and when-not-to-use rules: call again after processing done-annotations, do not call again when status is closed (user-initiated) or done with empty annotations, and only ask the user about reopening for window_lost/agent_cancelled/agent_exited. It also instructs passing session_id into sibling tools and when to set timeout_seconds for client-limited calls. This is textbook-level usage guidance with explicit exclusions and alternatives.

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

• annotate_documentA
  4.6/5.0
  Behavior5/5

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

  With no annotations provided, the description carries the full behavioral burden and does so thoroughly: it discloses session creation, local HTTP service startup, window reuse/in-place reload, clearing of previous annotations, automatic reload after apply_edit via server push, and the returned {session_id, url}. This goes well beyond what the schema alone communicates.

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

  Conciseness4/5

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

  The description is information-dense and front-loads the primary action and return value before the multi-turn loop explanation. Minor redundancy exists: the final sentence about reusing existing windows repeats the earlier reuse statement, slightly padding an otherwise efficient description.

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

  Completeness5/5

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

  Despite having no output schema, the description states the return value explicitly. It also covers the window lifecycle, the automatic reload mechanism, and the exact conditions under which the agent should call the tool again. For a tool that starts a local service and manages a browser window, nothing needed for a correct call is missing.

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

  Parameters4/5

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

  Schema coverage is 100% and the schema already describes path as an absolute .docx/.xlsx/.pptx path. The description adds meaning by making path the identity used for window reuse ('同一文档'), which clarifies that the same path maps to the same session/window. That is value beyond the schema's type-level description.

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

  Purpose5/5

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

  The description names a specific action ('打开文档标注页面'), explains the underlying work (create annotation session, start local HTTP service, open an H5 page in a standalone browser window), and distinguishes the tool's key reuse behavior from a simple one-shot opener. It leaves no ambiguity about the resource it operates on.

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

  Usage Guidelines4/5

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

  It provides clear when-to-use and when-not-to-use guidance: call at the start of a round or when an immediate refresh is needed, but not after every edit because the window auto-reloads after apply_edit. It does not explicitly name alternatives among the sibling tools, but the behavioral loop guidance is strong enough to route an agent correctly.

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

• apply_editA
  4.7/5.0
  Behavior5/5

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

  With no annotations provided, the description fully discloses the mutation behavior: automatic .bak-<timestamp> backup before first edit, version snapshot only on the first edit of a round, later edits reusing that snapshot, and version.desc representing the pre-edit state. It also discloses structured engine errors, the exact return shape, the auto-reload side effect, and a rollback path.

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

  Conciseness5/5

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

  The description front-loads the core purpose in the first clause and every subsequent sentence adds a distinct operational fact: backup generation, version collapsing, return values, auto-reload behavior, and rollback. It is dense but contains no filler or repeated schema information.

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

  Completeness5/5

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

  For a mutating tool with six parameters, nested objects, and no output schema, the description defines the return tuple, version semantics, backup and rollback path, and follow-up workflow with wait_for_annotations. No critical input needed to call the tool or react to its result is missing; only the exact structured-error shape is summarized rather than enumerated.

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

  Parameters4/5

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

  Schema description coverage is 100%, so the baseline is 3; the description adds meaningful semantics beyond the schema: mode defaults to replace, delete ignores new_content, and session_id causes same-round edits to merge into one version snapshot. It also explains the meaning of version and version.desc in the response, which the input schema does not cover.

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

  Purpose5/5

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

  The description states a specific operation: '对指定位置应用修改' (apply modifications at a specified location), immediately clarifying that this is a location-targeted edit tool. It also distinguishes its workflow from annotate_document by stating that the annotation window reloads automatically, and the loc-kind list reinforces that this is structural-position editing rather than a template or search operation.

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

  Usage Guidelines4/5

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

  The description provides strong workflow guidance: call this after receiving annotations, do not manually call annotate_document afterward, call wait_for_annotations again for the next round, and use list_versions/restore_version if the edit result is wrong. It does not explicitly compare against template_replace or find_replace, so the selection boundary among sibling edit tools is left somewhat to inference.

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