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

The annotations already declare readOnlyHint=true and openWorldHint=false, and the description's 読み取りのみ is consistent with them rather than contradictory. It adds genuinely useful behavior beyond the annotations: the fixed category buckets and the caveat that entry sizes include their subtree, so parent and child sizes overlap. It says nothing about cost, performance, or traversal limits (e.g. symlinks, permission errors, unreadable dirs), which matters for a filesystem walk.

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?

Four short sentences with no filler: purpose, then payload contents, then the size-overlap caveat, then the drill-down workflow. The most decision-relevant facts (what it returns, and that sizes double-count across levels) are front-loaded.

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?

An output schema exists, so the description doesn't need to enumerate return fields, and the annotations carry the safety profile. Given only two parameters and high schema coverage, the description is nearly complete; the only gap is the absence of any note on error or edge behavior (inaccessible paths, symlinks, large trees).

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 both parameters are already documented in the schema (including max_depth as 1=直下の子のみ and target_path as an absolute path allowing ~). The description's mention of max_depth and of reusing a discovered path adds only light framing on top of that. Baseline 3 applies when the schema carries the parameter documentation.

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 concrete verb+resource: it returns a capacity breakdown (容量内訳) for a specified directory, and explicitly notes 読み取りのみ. It also states what the payload contains (カテゴリ別合計 dev_cache/video/log/other plus top entries by size). It does not name or contrast with its siblings (scan_cleanup_targets, get_os_env_paths), so differentiation is left to the reader.

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 final sentence gives concrete operational guidance: after finding a large path, call again with that path to drill down efficiently, describing an iterative workflow. It gives no explicit when-not-to-use rule and never names an alternative sibling tool for the cleanup/environment-inspection case.

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 only declare readOnlyHint and openWorldHint=false, so the description adds real value by disclosing two behavioral traits: results are OS-dependent and filtered to directories that actually exist on disk. It stops short of saying whether existence is checked live or what happens when nothing is found, so not a 5.

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?

Two short sentences with no filler: the behavior comes first and the usage condition second. Every clause carries information an agent needs.

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?

With no params, an output schema, and read-only annotations, the description only needs to convey behavior and trigger, which it does. It could add one line on what the returned paths are suitable for (candidates, not confirmed cleanup targets), but nothing critical 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?

The tool takes zero parameters and the description correctly says nothing about inputs, matching the empty schema. Baseline 4 applies for a no-argument tool with full schema 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 specific verb and resource: returns the representative cache/log directories that actually exist for the current OS. Scope (OS-dependent, existence-filtered, no arguments) is clear enough to separate it from scan_cleanup_targets, though no sibling is named explicitly.

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?

Explicitly states the triggering condition: use it to find candidate locations when the user has not specified what to clean. That gives a clear when-to-use, but it never says when not to use it (e.g. after targets are already known) or names the alternative tool.

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?

Beyond the annotations (destructiveHint=true), the description adds the crucial nuance that deletion is recoverable via the OS trash, mandates user consent before execution, and discloses that non-scan and protected paths are silently skipped. These are exactly the behavioral facts an agent cannot infer from the boolean hints.

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?

Four short lines, front-loaded with the action and its non-destructive nature, followed by the consent precondition and the skip rule. No redundant restatement of the tool name or schema.

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?

With an output schema present, rich annotations, and fully documented parameters, the description covers everything else an agent needs: recoverability, consent requirement, and skip semantics. Nothing material is missing for correct invocation.

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, but the description adds a filtering rule not in the schema: paths outside the scan result and protected paths are skipped rather than rejected. That changes how an agent should interpret partial success from the paths array.

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 (move to OS Trash/Recycle Bin) and immediately disambiguates it from permanent deletion ('完全削除はしない'), which is the key distinction an agent must make among cleanup-related siblings. The reference to scan results ties it to scan_cleanup_targets.

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 an explicit precondition ('必ずユーザーの明示的な同意を得てから実行すること') and implies the tool is used on paths produced by a prior scan. It does not name an alternative tool or describe a case where a different sibling should be chosen instead, so it falls short of a full when/when-not specification.

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 readOnlyHint=true and openWorldHint=false; the description reinforces this with an explicit 'Dry-Run, deletes nothing' statement rather than contradicting it. It also discloses the cross-tool contract (scan_id feeds move_to_trash) and the human-consent step, which annotations cannot convey. It does not cover scan duration or result volume, keeping it below 5.

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?

Three short sentences, front-loaded with what the tool returns and ordered by size, followed by the dry-run guarantee and the downstream requirement. No filler.

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?

An output schema exists so return-value explanation is unnecessary, and the description still flags the key output (scan_id) and the required consent step. What is missing is disambiguation from analyze_disk_usage for an agent choosing between the two scanning tools.

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%, including an enum explanation for target_type and per-field descriptions for size and day thresholds, so the schema carries the parameter burden. The description only gestures at 'conditions' without adding format or default guidance, which is the expected baseline when the schema is complete.

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 resource ('enumerate deletion candidates matching conditions, ordered by size descending') and immediately marks the operation as dry-run. It clearly separates itself from move_to_trash by noting nothing is deleted, though it does not articulate how it differs from the sibling analyze_disk_usage.

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 real workflow guidance: the returned scan_id is required by move_to_trash, and results should be presented to the user for consent. It stops short of naming when to prefer analyze_disk_usage or which target_type to pick, so selection among scan modes is left to the schema.

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