datagokr — 공공데이터포털 검색을 더 쉽게 (find data.go.kr datasets easily)

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

The annotations already declare readOnly, idempotent, openWorld, and non-destructive, so the baseline bar is lower. The description goes beyond them usefully: it states the server never stores files (the call returns instructions, not data), describes the two response shapes (url vs post_url/get_url), and discloses the deleted-ID fallback returning 'deleted·alternatives'. That is real behavioral context an agent could not get from the annotations.

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 front-loaded with the key constraint ('서버가 파일을 대신 저장하진 않고') and repeats it in English for emphasis. The return-field enumeration is redundant given the output schema, and the trilingual layering adds length, so a 4 is warranted rather than a 5.

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 one-parameter read tool with a rich annotation set and the presence of an output schema, the definition covers the essential points: no file storage, return shape, examples, and the deleted-ID edge case. It lacks an explicit statement of the dataset_id format or of how this relates to the sibling lookup tools. Those are minor omissions, not blockers.

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 0% and there is one parameter, so the description has to carry the meaning of dataset_id. It never defines the field explicitly, but the numeric example IDs ('15028200', '15000001') and the note that a deleted ID returns alternatives partially convey the expected format and lifecycle. This is helpful but short of a full explanation.

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 gives a specific verb and output shape: 'Get client-side download instructions (URL, params, steps)' and 'The server never stores files.' That distinguishes it from an actual file-fetch tool. It never names or contrasts the sibling tools (get_preview, show, record), so the differentiation is implicit at best.

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?

Usage can be inferred from the examples ('15028200 전체 JSON 받는 법') and from the statement that the server does not store files, which implies you call this when you need the download procedure, not the data. There is no explicit statement of when to use this instead of get_preview or show, and no stated preconditions.

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 cover safety (readOnly, idempotent, non-destructive), and the description adds real behavior: AND semantics across all listed columns, reverse-index matching, dtype/org narrowing, and n capped at 20. It does not explain what 'group' does or what happens for zero matches, so it is strong rather than complete.

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?

Content is dense and front-loaded, with the routing rule first and details second. The Korean/English duplication doubles the length, but each version is tight and every sentence carries 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 read-only search tool with an output schema (so return values need not be re-explained), the definition covers the filtering semantics, the query constraint, and the result cap. The unexplained 'group' parameter is the one remaining gap.

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 0%, so the description must carry the load. It explains names (all columns required), dtype/org narrowing, and n's maximum of 20, which is genuine added meaning — but the 'group' boolean parameter is never mentioned, leaving one of five parameters undocumented anywhere.

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 states a specific verb+resource ('Find datasets with ALL given columns') and immediately contrasts itself with the sibling search tool. It also names the distinguishing constraint (no topic, query fixed to empty string), so an agent can tell it apart from search 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?

Explicit when-to-use ('no topic', query pinned to empty string) and an explicit alternative ('for a topic use search(query, fields=...)'). Routing is unambiguous.

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 (readOnly, idempotent, openWorld, non-destructive), the description discloses authentication behavior (the caller's X-DataGoKr-Key may be needed for API fallback), a privacy guarantee (keys are never returned or stored), result caps (50 columns / 20 rows, 5 rows for Seoul/Gyeonggi public paths), and failure semantics (ok=false means the portal is not responding right now, retry later; deleted IDs return deleted and alternatives). That is genuine behavioral context the annotations cannot convey.

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 purpose is front-loaded in the first sentence, which is good. However, the Korean and English halves largely duplicate each other (including a repeated explanation of data/columns/rows and ok=false), which roughly doubles the length without adding information, and the return-value prose is redundant given an output schema exists.

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 tool whose behavior branches across many dataset types, this description covers the important branches, the key/auth caveat, the row/column caps, and error recovery. Since an output schema exists, the detailed return-field walkthrough is surplus rather than necessary, but nothing critical to calling it correctly 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 coverage is 0%, so the description carries the full burden for the two parameters. It implies what dataset_id means through references to deleted IDs and per-type routing, and it hints at row-count limits ('최대5행', '최대 20행'), but it never explains that 'n' is the requested row count or what its default of 5 governs relative to those caps. Partial compensation only.

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 opening sentence pairs a specific verb ('실제 첫 행 보기 / Preview') with a specific resource (first rows of datagokr datasets) and immediately scopes it by source type (STD_FILE, PORTAL_FILE, OPEN_API, LINK, STD child rows). That is enough to separate it from siblings like search, record, or download_url, though none of those siblings are named explicitly as alternatives.

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 gives conditional routing rules: OPEN_API datasets return request_templates, LINK types return url, STD child rows need parent_id, and a failed PORTAL_FILE attachment preview may fall back to an API call requiring the caller's own key. It also says what to do on ok=false (retry later) and on deleted IDs (deleted plus alternatives). What is missing is an explicit statement of when to prefer a sibling tool over this one.

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/idempotent/non-destructive/closed-world, so the safety profile is covered. The description adds genuinely new behavior: internal state is excluded, nothing is applied or stored server-side, and deleted IDs return a `deleted` marker plus `alternatives` — a non-obvious edge case an agent would otherwise miss.

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 return-field list is front-loaded and useful, but the entire description is duplicated verbatim in Korean and English, roughly doubling length without adding information. The content is sound; the packaging wastes half the space.

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, yet the description still usefully summarizes the shape and covers the deleted-ID fallback, which the schema likely cannot express. The only real omission is how to source a valid `dataset_id`, a small gap for a one-parameter read tool.

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 single parameter `dataset_id` has 0% schema description coverage, so the description must carry the load. It implicitly frames the ID as a dataset identifier and warns that a deleted ID yields `deleted`/`alternatives`, but never says where to obtain a valid ID (presumably from `search`/`show`) or its format.

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 raw metadata needed for 조회·신청·다운로드, scoped to the local datagokr package. It explicitly distinguishes itself from the sibling `show` tool, so an agent can choose between them 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 clear routing guidance ('일반 사용에서는 show를 쓰세요' / 'prefer `show` for normal use') and a scope restriction (local datagokr client only). The residual gap is that it never states the concrete condition under which the raw form is actually required, only that normal use is not it.

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 read-only, idempotent, open-world, non-destructive behavior, and the description adds real context beyond them: results are collapsed by topic by default via group=True, collapsed groups must be expanded through `show` using group_ids, n is capped at 20, and region_gap/region_fallback/scope_warning signals appear in the response. It loses a point because the closing '반환:' line largely restates fields already defined in the output schema.

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 purpose, then organized as scannable bullets covering the main pitfalls in priority order. Some bulk comes from Korean/English duplication of the purpose line and the redundant return-field listing.

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 6-parameter search tool with grouping and regional fallbacks, the description covers the full workflow: query formulation, filter usage, result collapsing, expansion via `show`, retry guidance, and coverage-gap signals. The output schema covers return shape, so no more is needed there.

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 0%, so the description carries the full burden and mostly does: `query` is a one-line topic phrasing with examples, `fields` holds column conditions, `dtype` accepts API/FILE, `org` filters by institution, `group` toggles collapsing, and `n` is capped at 20. It does not explain matching semantics for `org`/`dtype` (exact vs partial, case sensitivity), leaving a small gap.

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 ('주제로 공공데이터를 찾습니다 / Search Korean public datasets by topic') and scopes it to the datagokr portal. It also distinguishes itself from the sibling `fields` tool for column-only conditions, so an agent can route 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 explicit when-to-use routing: topic+column conditions should pass `fields` alongside the query, while column-only conditions belong to the `fields` tool. It also prescribes fallback behavior (drop filters when results look wrong, use region_fallback when no local data exists, split evaluation-style asks into 2-3 searches) and mandates follow-up with `show`.

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 readOnly/idempotent/non-destructive annotations, the description discloses truncation behavior (responses capped near 20KB, columns capped at 50, version_keys limited to the newest 3), the truncated/more fields that let the agent recover cut data, and the deleted/alternatives path for removed IDs. These are exactly the behavioral traits an agent cannot infer from structured fields.

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 dense but well front-loaded: purpose first, then the return payload, then the truncation caveat, then the operation narrowing and next-step pointer. The long enumeration of return fields is somewhat redundant given an output schema exists, but it does double as disclosure of the truncation surface.

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 two-parameter read tool with full annotations and an output schema, the description covers the remaining unknowns well: response size limits, how to drill into one operation, what happens for deleted IDs, and the follow-up with get_preview. Only the id vs dataset_id naming slip and the unaddressed fields/record overlap keep it short of complete.

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 0%, so the description carries the burden and mostly does: it explains that operation accepts a function name or operation_seq and is used to view a single operation in full, and dataset_id is implicitly the dataset selected from search. The only blemish is that it writes 'show(id, ...)' while the actual parameter is dataset_id, a small naming mismatch.

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 gives a concrete verb and resource: view the structure and access methods of a dataset chosen from the datagokr search, and it names get_preview as the tool for actual row data. It does not, however, differentiate itself from siblings like fields or record, which sound like they could also return schema-like information, so the agent must guess at that boundary.

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?

It states a clear next-step alternative ('if it looks usable, check real rows with get_preview') and explains how to narrow an oversized API response via the operation parameter. It also covers the deleted-ID case with alternatives. What is missing is explicit guidance on when to use show instead of fields or record.

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