> For clean Markdown of any page, append .md to the page URL. > For a complete documentation index, see https://docs.meetstream.ai/llms.txt. > For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.meetstream.ai/_mcp/server. # MeetStream Guide: Amazon S3 Custom Object Storage By default, MeetStream stores bot media—audio, video, transcripts, screenshots, chat logs, and meeting manifests—in a MeetStream-managed bucket. Fetch endpoints return presigned download URLs for those files. Custom object storage lets you send those artifacts directly to your own storage account while keeping the same MeetStream bot and fetch APIs. Use the `aws` provider for Amazon S3 or another S3-compatible endpoint. Authenticate with an AWS access key pair or IAM role. Only one S3-compatible custom provider is active at a time. Saving a new `aws` or `alibaba_oss` configuration activates it and deactivates the other one. Artifacts written before the change remain pinned to their original provider, bucket, and prefix. Custom storage is useful for compliance, regional data storage, lifecycle control, or direct ingestion into your data platform. --- ## 1) Prerequisites All providers require: 1. A bucket owned by your organization. 2. Credentials that can write to the prefix you configure. 3. A MeetStream API key. For `read_write` access, the credentials also need to read uploaded objects so MeetStream can generate download URLs and retrieve JSON artifacts. Create an IAM user access key or an IAM role that has access to your bucket and configured prefix. --- ## 2) Configure custom storage ```http PUT /api/v1/admin/configs?config_type=storage Authorization: Token Content-Type: application/json ``` ```json { "provider": "aws", "bucket_name": "your-s3-bucket", "region": "us-east-1", "access_key_id": "AKIA...", "secret_key": "...", "access_mode": "read_write", "prefix": "meetstream", "prefixes": { "audio": "recordings/audio", "video": "recordings/video", "transcript": "recordings/transcripts", "metadata": "recordings/metadata" } } ``` ### Parameters | Parameter | Required | Description | |---|---:|---| | `provider` | Yes | `"aws"`. | | `bucket_name` | Yes | Name of your S3 bucket. | | `region` | Yes | AWS region. | | `access_key_id` | Conditional | Required with `auth_method: "access_keys"`. | | `secret_key` | Conditional | Required with `auth_method: "access_keys"`. | | `auth_method` | No | `"access_keys"` (default) or `"iam_role"`. | | `role_arn` | Conditional | Required only with `auth_method: "iam_role"`. | | `external_id` | Conditional | Required only with `auth_method: "iam_role"`. | | `access_mode` | No | `"read_write"` (default) or `"write_only"`. See [Access modes](#4-access-modes). | | `prefix` | No | Base object-key prefix. Defaults to `"meetstream"`. Leading/trailing slashes are removed and `..` is rejected. | | `prefixes` | No | Per-category overrides for `audio`, `video`, `transcript`, and `metadata`. A missing category inherits `prefix`. | | `endpoint_url` | No | HTTPS endpoint override that can target another S3-compatible endpoint. | Secrets are stored as encrypted parameters and are never returned by the API. --- ## 3) Grant bucket permissions Scope policies to the configured prefix. If you use distinct category prefixes, include one object resource for each distinct prefix. For a `read_write` configuration using the `meetstream` prefix: ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Allow", "Action": "s3:ListBucket", "Resource": "arn:aws:s3:::your-s3-bucket" }, { "Effect": "Allow", "Action": [ "s3:PutObject", "s3:GetObject", "s3:DeleteObject" ], "Resource": "arn:aws:s3:::your-s3-bucket/meetstream/*" } ] } ``` For `write_only`, remove `s3:ListBucket` and `s3:GetObject`; keep `s3:PutObject` and `s3:DeleteObject`. --- ## 4) Access modes | Mode | MeetStream uploads to your bucket | MeetStream fetch APIs | |---|---|---| | `read_write` (default) | Yes | Return presigned URLs or read JSON artifacts from your bucket | | `write_only` | Yes | Return `403`; retrieve objects directly through your own storage access | Use `write_only` when you want MeetStream to deliver files to your bucket but do not want it to read them back. ### Configuration validation MeetStream validates credentials before saving them: - `aws` with `read_write` validates bucket access with `HeadBucket`. - `write_only` configurations use temporary write probes under each distinct effective prefix. Successful probes are cleaned up. Grant `DeleteObject` to prevent a failed cleanup from leaving a probe object behind. --- ## 5) File layout in your bucket MeetStream writes every artifact below the resolved prefix for its category, followed by the bot ID: ```text // ``` With the default `prefix: "meetstream"` and no per-category overrides: ```text meetstream/ abc123/ audio.wav meeting_recording.mp4 manifest.json participants.json chats.json screenshot/ transcription/ ``` The exact provider, bucket, and category prefix used for each processed artifact are recorded with the bot or transcript. Updating a storage configuration later does not break retrieval of existing media. --- ## 6) Fetch media after configuration Your MeetStream API calls do not change when custom storage is enabled. With `read_write`, the API returns a presigned URL generated by the configured provider. | Endpoint | `read_write` | `write_only` | |---|---|---| | `GET /api/v1/bots/{bot_id}/get_audio` | Presigned mixed-audio URL | `403` | | `GET /api/v1/bots/{bot_id}/get_video` | Presigned recording URL | `403` | | `GET /api/v1/bots/{bot_id}/get_audio_streams` | Per-participant audio URLs, when requested for the bot | `403` | | `GET /api/v1/bots/{bot_id}/get_recording_streams` | Per-participant video URLs, when requested for the bot | `403` | | Screenshot, chat, participant, manifest, and transcript fetch endpoints | Read from the configured bucket | `403` | Presigned media URLs expire after one hour. Call the fetch endpoint again to obtain a fresh URL. --- ## 7) View, change, or delete configuration ### View ```http GET /api/v1/admin/configs Authorization: Token ``` The response includes non-sensitive metadata under `StorageConfig`. ### Delete ```http DELETE /api/v1/admin/configs?key_name=aws Authorization: Token ``` Deleting a configuration affects only future processing; MeetStream does not delete objects already written to your bucket. --- ## 8) Troubleshooting **A fetch endpoint returns `403` with `"access_mode": "write_only"`** The configured provider is intentionally write-only. Download the object through your own storage workflow, or update it to `read_write` with read permissions. **A presigned URL returns `403`** - It may have expired; fetch a new one from MeetStream. - Verify the configured credentials still have object read access. **Files are not in the newly configured bucket** Only artifacts processed after a configuration change use the new active provider. Existing artifacts remain in their original bucket.