Read, write, and list text objects under a scoped bucket prefix.
Read and write files in an S3 bucket, scoped to a key prefix.
Connect
| Credential field | Required | Where it comes from |
|---|
Access key ID (ACCESS_KEY_ID) | Yes | |
Secret access key (SECRET_ACCESS_KEY) | Yes | |
| Setting | Required | What it is |
|---|
Bucket (BUCKET) | Yes | |
Key prefix (PREFIX) | No | Agents can only touch keys under this prefix |
Region (REGION) | No | e.g. us-east-1 |
Endpoint override (ENDPOINT) | No | For S3-compatible stores; leave empty for AWS. Must be a public https host. |
| Tool | Access | What it does |
|---|
Put object (yekar.s3.put_object) | Write | Write a text object under the scoped prefix (idempotent: same key + same body re-runs safely) |
Get object (yekar.s3.get_object) | Read | Read an object under the scoped prefix. Small text objects come back inline as body; anything large or binary (an image, a PDF, an archive) is written to a file instead and returned as an attachment. |
List objects (yekar.s3.list) | Read | List keys under the scoped prefix |
Notes
- Agents can only touch keys under the prefix you configure - “..” and out-of-prefix keys are rejected.
- An access key ID and secret access key are required.
- Set the endpoint override to use an S3-compatible store (MinIO, localstack). It must be a public https host (http is allowed only for localhost).