nsfw_detectionDetect whether an image contains NSFW (violent or sexually explicit) content and return an nsfw_score. 이미지가 NSFW(폭력적·선정적) 콘텐츠인지 탐지해 nsfw_score 를 반환합니다. detail=1 입력 시 세부 판정 결과를 함께 반환합니다. [호출당 10포인트]
| Name | Required | Description | Default |
|---|---|---|---|
| detail | No | 세부 판정 결과 포함 여부 (포함: 1, 미포함: 0, 기본값 0) | |
| image_url | Yes | 다운로드 가능한 https URL (허용 형식: image/jpeg, image/png, image/webp, image/bmp) (최대 50MB) |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint=true, so the read-only nature is covered. The description adds valuable context beyond that: the cost note ('[호출당 10포인트]' – 10 points per call) and the behavior of the detail parameter (returns detailed results when set to 1). These add practical constraints and outcome expectations beyond the annotation-provided safety profile.
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 compact and front-loaded with the core purpose in the first sentence. However, it is bilingual (English and Korean), repeating the same information twice, which adds some redundancy. The cost note is placed at the end, which is acceptable. Overall, it is concise enough and earns a 4.
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?
The description covers the core purpose, the detail parameter, and the cost, but it omits details about the return format (e.g., nsfw_score range, thresholds, or how the score should be interpreted). Since there is no output schema, the agent lacks guidance on what the returned value means or how to handle it. This is a notable gap for a tool intended to be called autonomously.
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%, so both parameters (image_url and detail) are fully documented in the schema. The description merely reiterates that detail=1 returns additional detail, which is already in the schema. It adds no new semantic meaning, so it meets the baseline for high schema coverage without compensating further.
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 clearly states the tool's function: detect NSFW (violent or sexually explicit) content in an image and return an nsfw_score. It uses a specific verb ('Detect') and names the resource ('image') and outcome ('nsfw_score'). It also distinguishes itself from sibling tools like face_detection or image_similarity by focusing solely on NSFW classification.
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 implies when to use this tool (when an image needs NSFW screening) but does not explicitly state alternatives or when-not-to-use scenarios. Since there is no other NSFW tool among siblings, the purpose is self-evident, but explicit guidance is absent, so it only meets the 'implied usage' threshold.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Add one secure layer between your agents and this server.
Most tools have clearly distinct purposes; even within families like identi_card1-5 vs identi_card_image1-5, the text-input vs image-input distinction is clear. However, the sheer number of tools and some near-synonyms (e.g., ocr_identi1 vs identi_card_image1) could cause occasional misselection, but descriptions mitigate this.
Naming follows a loose verb-first pattern (check_, crawl_, download_, draw_, etc.) but includes significant deviations: bare nouns (bank_code, location, whois), numbered variants (identi_card1, identi_card_image1), and mixed prefixes (ocr_, identity_, etc.). The inconsistency is noticeable but still readable and predictable within functional clusters.
80 tools is far above the typical 3-15, but the server is a broad API aggregator covering many independent domains (banking, ID verification, media conversion, search, LLM, etc.), so the high count is somewhat justified. Still, the sheer number makes the toolkit feel unwieldy and hard to navigate, placing it at the high end of acceptable.
Within its stated purpose as a general-purpose utility API, the toolset covers a wide array of common task families: identity document verification (text and image), OCR field extraction, media conversion, web/search, domain/IP lookup, and LLM chat. Most operations have both get and act variants (e.g., set/get watermark, parcel_tracking/auto), with few obvious dead ends for typical use cases.