> ## Documentation Index > Fetch the complete documentation index at: https://docs.bodyguard.ai/llms.txt > Use this file to discover all available pages before exploring further. # Analyze an image > Analyzes an image by URL or by inline image bytes. Returns a list of classifications for the image and, when OCR is enabled, the text extracted from the image along with its own text analysis. Only **PNG**, **JPEG** and **WebP** images are accepted, and the image must not exceed **1 MiB** (1,048,576 bytes). Requests with an unsupported format or an oversized image are rejected with a `400 Bad Request`. Before diving in, we recommend reading the [image taxonomy](/en/documentation/taxonomy/image-classifications) to understand the classifications returned by this endpoint. ## OpenAPI ````yaml /openapi/analyze.openapi.yaml post /analyze/v1/image openapi: 3.1.0 info: title: Bodyguard Analyze API version: 1.0.0 servers: - url: https://api.bodyguard.ai security: - ApiKeyAuth: [] tags: - name: Analyze description: Content analysis endpoints. paths: /analyze/v1/image: post: tags: - Analyze summary: Analyze an image description: > Analyzes an image by URL or by inline image bytes. Returns a list of classifications for the image and, when OCR is enabled, the text extracted from the image along with its own text analysis. Only **PNG**, **JPEG** and **WebP** images are accepted, and the image must not exceed **1 MiB** (1,048,576 bytes). Requests with an unsupported format or an oversized image are rejected with a `400 Bad Request`. Before diving in, we recommend reading the [image taxonomy](/en/documentation/taxonomy/image-classifications) to understand the classifications returned by this endpoint. operationId: bodyguard.analyze.v1.AnalyzeImage requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/bodyguard.analyze.v1.AnalyzeImageRequest' examples: byUrl: summary: Analyze an image by URL value: channelId: __YOUR_CHANNEL_ID__ publishedAt: '2026-04-23T10:30:00.000Z' imageUrl: https://cdn.example.com/uploads/photo.jpg multipart/form-data: schema: $ref: '#/components/schemas/bodyguard.analyze.v1.AnalyzeImageRequest' examples: byBytes: summary: Analyze a base64-encoded image value: channelId: __YOUR_CHANNEL_ID__ imageData: '[file content goes there]' responses: '200': description: Analysis completed. content: application/json: schema: $ref: '#/components/schemas/bodyguard.analyze.v1.AnalyzeImageResponse' examples: classified: summary: Image classified as sexual content value: imageAnalysis: classifications: - PARTIAL_NUDITY classificationMeta: PARTIAL_NUDITY: confidenceScore: 0.92 ocrAnalysis: text: Text extracted from the image textAnalysis: type: HATEFUL classifications: - INSULT severity: HIGH directedAt: AUTHOR_OF_COMMENT language: en recommendedAction: REMOVE censoredText: Einstein is an ******* classificationMeta: INSULT: severity: MEDIUM components: schemas: bodyguard.analyze.v1.AnalyzeImageRequest: type: object required: - channelId properties: channelId: type: string description: >- Identifier of the [channel](/en/api-reference/concepts#channel) the image belongs to. Provided by Bodyguard. publishedAt: type: string format: date-time description: >- [Publication date](/en/api-reference/concepts#publication-date-publishedat) of the image. imageUrl: type: string format: uri description: > Public URL of the image to analyze. The image must be a PNG, JPEG or WebP file no larger than 1 MiB. Only one of `imageUrl` or `imageData` can be provided. When fetching the image, Bodyguard sends the user-agent `Bodyguard-Analyze/1.0 (+https://bodyguard.ai/bot)`. If you allowlist Bodyguard on a CDN or firewall, match on the stable `Bodyguard-Analyze` product token rather than the full string — the version number and bot URL may change over time. Image fetches are not sent from a fixed set of IP addresses, so the user-agent is the signal to allowlist against. imageData: type: string format: byte description: >- Base64-encoded image bytes. The image must be a PNG, JPEG or WebP file no larger than 1 MiB. Only one of `imageUrl` or `imageData` can be provided. oneOf: - required: - imageUrl - required: - imageData bodyguard.analyze.v1.AnalyzeImageResponse: type: object properties: imageAnalysis: $ref: '#/components/schemas/bodyguard.analyze.v1.ImageAnalysis' bodyguard.analyze.v1.ImageAnalysis: type: object description: Moderation result for an image. properties: classifications: type: array description: >- Classifications detected on the image itself. An empty array means the image was analysed successfully and Bodyguard did not detect any sensitive content — it can be treated as safe. items: type: string classificationMeta: type: object additionalProperties: $ref: >- #/components/schemas/bodyguard.analyze.v1.ImageAnalysis.ClassificationMeta ocrAnalysis: description: >- Text extracted from the image by OCR, together with a text analysis. `null` if OCR is disabled or no text was found in the image. oneOf: - $ref: >- #/components/schemas/bodyguard.analyze.v1.ImageAnalysis.OcrAnalysis - type: 'null' bodyguard.analyze.v1.ImageAnalysis.ClassificationMeta: type: object description: >- Additional metadata for each detected image classification. Keys match entries in `classifications`. properties: confidenceScore: type: number format: float minimum: 0 maximum: 1 description: >- Bodyguard's confidence that the classification applies to the image, between `0` and `1`. Higher means more confident. Relying on this score is discouraged: Bodyguard already filters out classifications whose confidence is too low, so the presence of a classification in the response is the signal you should use to decide whether to act on it. bodyguard.analyze.v1.ImageAnalysis.OcrAnalysis: type: object properties: text: type: string description: The text snippet extracted from the image. textAnalysis: $ref: '#/components/schemas/bodyguard.analyze.v1.TextAnalysis' bodyguard.analyze.v1.TextAnalysis: type: object description: Moderation result for a piece of text. properties: type: $ref: '#/components/schemas/bodyguard.std.v1.Type' classifications: type: array description: >- Fine-grained labels describing the content (e.g. `INSULT`, `SPAM`). See the [text taxonomy](/en/documentation/taxonomy/text-classifications) for the full list and descriptions of each classification. items: type: string severity: $ref: '#/components/schemas/bodyguard.std.v1.Severity' directedAt: $ref: '#/components/schemas/bodyguard.std.v1.DirectedAt' language: type: string description: Detected language as an ISO 639-1 code. recommendedAction: $ref: '#/components/schemas/bodyguard.std.v1.RecommendedAction' censoredText: type: string description: The original text with sensitive content censored with `*`. classificationMeta: type: object additionalProperties: $ref: >- #/components/schemas/bodyguard.analyze.v1.TextAnalysis.ClassificationMeta bodyguard.std.v1.Type: type: string description: >- The broad category of the analyzed content. For finer-grained detail, see `classifications`. See the [text taxonomy](/en/documentation/taxonomy/text-classifications) for the full description of each value. enum: - NEUTRAL - HATEFUL - POSITIVE - NONE - HATE_SPEECH - CRITICISM - UNDESIRABLE bodyguard.std.v1.Severity: type: string description: >- Severity of the detected issue, if any. See the [text taxonomy](/en/documentation/taxonomy/text-classifications) for the full description of each value. enum: - LOW - MEDIUM - HIGH - CRITICAL - NONE bodyguard.std.v1.DirectedAt: type: string description: >- Who or what the content is directed at. See the [text taxonomy](/en/documentation/taxonomy/text-classifications) for the full description of each value. enum: - USER - USER_FAMILY - SINGLE_PERSON - GROUP - EVERYONE - AUTHOR_OF_COMMENT - HATERS - NONE - ENTITY bodyguard.std.v1.RecommendedAction: type: string description: The moderation action Bodyguard recommends you apply to this content. enum: - KEEP - REMOVE - WATCH bodyguard.analyze.v1.TextAnalysis.ClassificationMeta: type: object description: > Additional metadata for each detected classification. **Important:** Bodyguard cannot guarantee that this metadata will be present for every classification — for example, some non-toxic classifications may not have any." properties: severity: $ref: '#/components/schemas/bodyguard.std.v1.Severity' securitySchemes: ApiKeyAuth: type: apiKey in: header name: Authorization description: | API key authentication. Provide your key in the `Authorization` header with the `ApiKey` scheme: ``` Authorization: ApiKey ``` ````