Open-source platform for 3D AI agents. Turn text or a photo into a rigged, animated GLB avatar, give it an LLM brain, memory and a wallet, and embed it anywhere with one web component. Remote MCP server, x402 payments, Three.js, browser-native.
JavaScript 3d-generationai-agentsanimationavatardigital-humangenerative-aiglbgltfimage-to-3dllmmcpmcp-servermodel-context-protocolsolanatext-to-3dthreejsweb-componentswebglwebxrx402
/**
* Filebase S3-compatible IPFS pinning backend.
*
* Filebase exposes an S3 API at https://s3.filebase.com.
* After a PutObject, the IPFS CID is available via HeadObject in
* the x-amz-meta-cid response header.
*
* The @aws-sdk/client-s3 is dynamically imported so non-Filebase users
* do not pay the bundle cost.
*
* pinDirectory: uploads each file under a common key prefix, then returns the
* CID of the first file as a best-effort root. Note: Filebase does not expose
* a true directory-wrapped IPFS CID via S3; for proper directory CIDs use Pinata.
*/
import { withRetry, statusError } from './_retry.js';
export class FilebasePinner {
/**
* @param {object} opts
* @param {string} opts.accessKeyId Filebase S3 access key ID
* @param {string} opts.secretAccessKey Filebase S3 secret access key
* @param {string} opts.bucket Target bucket name
* @param {string} [opts.region='us-east-1']
*/
constructor({ accessKeyId, secretAccessKey, bucket, region = 'us-east-1' } = {}) {
this._creds = { accessKeyId, secretAccessKey };
this._bucket = bucket;
this._region = region;
this._client = null;
}
async _getClient() {
if (this._client) return this._client;
const { S3Client } = await import('@aws-sdk/client-s3');
this._client = new S3Client({
region: this._region,
endpoint: 'https://s3.filebase.com',
credentials: this._creds,
// Path-style required by Filebase
forcePathStyle: true,
});
return this._client;
}
/**
* @param {Blob|Uint8Array} blob
* @param {{name?: string, onProgress?: function(number): void}=} opts
* @returns {Promise<{cid: string, size: number}>}
*/
async pinBlob(blob, opts = {}) {
const { name = `upload-${Date.now()}` } = opts;
const { PutObjectCommand, HeadObjectCommand } = await import('@aws-sdk/client-s3');
const client = await this._getClient();
const body = blob instanceof Uint8Array ? blob : new Uint8Array(await blob.arrayBuffer());
const size = body.length;
await withRetry(async () => {
const res = await client.send(
new PutObjectCommand({
Bucket: this._bucket,
Key: name,
Body: body,
ContentType: blob instanceof Blob ? blob.type : 'application/octet-stream',
}),
);
if (res.$metadata?.httpStatusCode >= 500) {
throw statusError(res.$metadata.httpStatusCode, 'Filebase PutObject failed');
}
});
const head = await withRetry(async () => {
return client.send(new HeadObjectCommand({ Bucket: this._bucket, Key: name }));
});
const cid = head.Metadata?.cid;
if (!cid) throw new Error('Filebase did not return an IPFS CID in x-amz-meta-cid');
return { cid, size };
}
/**
* Uploads each file under a shared prefix. Returns the CID of the first file.
* Not a true directory-wrapped CID — use Pinata for {cid}/{path} resolution.
*
* @param {Array<{path: string, data: Blob|Uint8Array}>} files
* @param {{name?: string, onProgress?: function(number): void}=} opts
* @returns {Promise<{cid: string, size: number}>}
*/
async pinDirectory(files, opts = {}) {
const { name = `dir-${Date.now()}` } = opts;
let firstCid = null;
let totalSize = 0;
for (const { path, data } of files) {
const key = `${name}/${path}`;
const { cid, size } = await this.pinBlob(data, { name: key });
totalSize += size;
if (!firstCid) firstCid = cid;
}
if (!firstCid) throw new Error('No files provided to pinDirectory');
return { cid: firstCid, size: totalSize };
}
/**
* Delete the S3 object — Filebase will eventually GC the underlying IPFS pin.
* @param {string} key The S3 key (not CID) used at upload time
*/
async unpin(key) {
const { DeleteObjectCommand } = await import('@aws-sdk/client-s3');
const client = await this._getClient();
await withRetry(() =>
client.send(new DeleteObjectCommand({ Bucket: this._bucket, Key: key })),
);
}
}