| name | bkend-storage |
| classification | capability |
| classification-reason | Pattern guidance may overlap with model's built-in knowledge as it improves |
| deprecation-risk | medium |
| effort | medium |
| description | bkend.ai file storage — upload (presigned URL), download (CDN), visibility levels, buckets.
Triggers: bkend file, bkend upload, bkend download, bkend storage, bkend presigned URL
|
| user-invocable | false |
| agent | bkit:bkend-expert |
| allowed-tools | ["Read","Write","Edit","Glob","Grep","Bash","mcp__bkend__*"] |
| imports | ["${PLUGIN_ROOT}/templates/shared/bkend-patterns.md"] |
bkend.ai Storage Guide
Upload Methods
| Method | Use Case | Process |
|---|
| Single | Normal files | Presigned URL -> PUT upload -> Register metadata |
| Multiple | Multiple files | Repeat single upload |
| Multipart | Large files | Initialize -> Part URLs -> Complete |
Presigned URL
- Validity: 15 minutes
- PUT method with file binary
- Content-Type header required
File Visibility (4 levels)
| Level | Access | URL Type |
|---|
| public | Anyone | CDN URL (no expiry) |
| private | Owner only | Presigned URL (1 hour) |
| protected | Authenticated users | Presigned URL (1 hour) |
| shared | Specified targets | Presigned URL (1 hour) |
Size Limits
| Category | Max Size |
|---|
| Images | 10 MB |
| Videos | 100 MB |
| Documents | 20 MB |
Storage Categories
images, documents, media, attachments
MCP Storage Workflow
bkend MCP does NOT have dedicated storage tools. Use this workflow:
- Search docs:
search_docs with query "file upload presigned url"
- Get examples:
search_docs with query "file upload code examples"
- Generate code: AI generates REST API code for file operations
Searchable Storage Docs
| Doc ID | Content |
|---|
7_code_examples_data | CRUD + file upload code examples |
REST Storage API
| Method | Endpoint | Description |
|---|
| POST | /v1/files/presigned-url | Generate presigned URL |
| POST | /v1/files | Register metadata (complete upload) |
| GET | /v1/files | File list |
| GET | /v1/files/:fileId | File detail |
| PATCH | /v1/files/:fileId | Update metadata |
| DELETE | /v1/files/:fileId | Delete file |
| POST | /v1/files/:fileId/download-url | Generate download URL |
Multipart Upload (Large Files)
| Method | Endpoint | Description |
|---|
| POST | /v1/files/multipart/init | Initialize multipart upload |
| POST | /v1/files/multipart/presigned-url | Get part upload URL |
| POST | /v1/files/multipart/complete | Complete multipart upload |
| POST | /v1/files/multipart/abort | Abort multipart upload |
Upload Flow (Single File)
1. POST /v1/files/presigned-url -> { url, fileId }
2. PUT {url} with file binary + Content-Type header
3. POST /v1/files with { fileId, filename, contentType, size, visibility }
Multipart Upload Flow (Large File)
1. POST /v1/files/multipart/init -> { uploadId }
2. POST /v1/files/multipart/presigned-url -> [{ partNumber, url }]
3. PUT each part URL with file chunk
4. POST /v1/files/multipart/complete -> { file }
Official Documentation (Live Reference)
For the latest storage documentation, use WebFetch:
