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
import { sql } from '../../_lib/db.js';
import { crawlAgentAttestations, KIND_MAP } from '../../_lib/solana-attestations.js';
import { env } from '../../_lib/env.js';
async function ensureWarm({ asset, network }) {
const [c] = await sql`select 1 from solana_attestations_cursor where agent_asset = ${asset} limit 1`;
if (c) return;
const [agent] = await sql`
select wallet_address as owner from agent_identities
where meta->>'sol_mint_address' = ${asset} and deleted_at is null limit 1
`;
try { await crawlAgentAttestations({ agentAsset: asset, network, ownerWallet: agent?.owner || null }); }
catch { /* return cached (empty) */ }
}
// Exported so other paid surfaces (e.g. the persona identity/wallet tools) can
// reuse the SAME ERC-8004-style reputation computation instead of re-deriving
// it against a second copy of the attestation query.
export async function solanaReputation(asset, network) {
await ensureWarm({ asset, network });
const [fb] = await sql`
with feedback as (
select
f.disputed, (f.payload->>'score')::int as score,
exists (
select 1 from solana_attestations a
where a.agent_asset = f.agent_asset and a.kind='threews.accept.v1'
and a.payload->>'task_id' = f.payload->>'task_id'
and a.verified = true and f.payload->>'task_id' is not null
) as task_accepted
from solana_attestations f
where f.agent_asset=${asset} and f.network=${network}
and f.kind='threews.feedback.v1' and f.revoked=false
)
select count(*)::int total,
count(*) filter (where task_accepted)::int verified,
count(*) filter (where disputed)::int disputed,
coalesce(avg(score),0)::float score_avg,
coalesce(avg(score) filter (where task_accepted),0)::float score_avg_verified
from feedback
`;
const [val] = await sql`
select count(*) filter (where (payload->>'passed')::bool)::int passed,
count(*) filter (where not (payload->>'passed')::bool)::int failed
from solana_attestations
where agent_asset=${asset} and network=${network}
and kind='threews.validation.v1' and revoked=false
`;
return {
agent: asset, network,
feedback: { ...fb,
score_avg: Number(fb.score_avg.toFixed(3)),
score_avg_verified: Number(fb.score_avg_verified.toFixed(3)) },
validation: val,
};
}
async function solanaAttestations({ asset, kind, network, limit }) {
await ensureWarm({ asset, network });
const wantKind = kind === 'all' ? null : KIND_MAP[kind];
const rows = wantKind
? await sql`
select signature, slot, block_time, attester, kind, payload, verified, revoked, disputed
from solana_attestations
where agent_asset=${asset} and network=${network} and kind=${wantKind} and revoked=false
order by slot desc limit ${limit}
`
: await sql`
select signature, slot, block_time, attester, kind, payload, verified, revoked, disputed
from solana_attestations
where agent_asset=${asset} and network=${network} and revoked=false
order by slot desc limit ${limit}
`;
return { agent: asset, network, kind, count: rows.length, data: rows };
}
async function solanaPassport(asset, network) {
const [agent] = await sql`
select id, name, description, wallet_address as owner, meta
from agent_identities
where meta->>'sol_mint_address' = ${asset} and deleted_at is null limit 1
`;
const reputation = await solanaReputation(asset, network);
const recent = await solanaAttestations({ asset, kind: 'all', network, limit: 10 });
return {
agent: asset,
identity: agent ? {
id: agent.id, name: agent.name, description: agent.description,
owner: agent.owner, asset_pubkey: asset,
network: agent.meta?.network || network,
} : { agent_off_index: true, asset_pubkey: asset, network },
reputation: reputation.feedback,
validation: reputation.validation,
recent_attestations: recent.data,
schemas_url: `${env.APP_ORIGIN}/.well-known/agent-attestation-schemas`,
};
}
// MCP tool annotations (2025-06-18 spec): public, read-only views over the
// live on-chain attestation index. New attestations land between calls, so
// not idempotent. destructiveHint defaults to TRUE when omitted, so it is set
// explicitly.
const ATTESTATION_READ_ANNOTATIONS = {
readOnlyHint: true,
destructiveHint: false,
idempotentHint: false,
openWorldHint: true,
};
export const toolDefs = [
{
name: 'solana_agent_reputation',
title: 'Get Solana agent reputation',
annotations: ATTESTATION_READ_ANNOTATIONS,
description:
'Computed reputation summary for a Solana-registered three.ws agent. Returns total/verified feedback counts, score averages (raw + verified-only), validation pass/fail, task acceptance, and dispute counts. Verified score only includes feedback whose task was acknowledged on-chain by the agent owner. Priced at $0.001 USDC/call via x402 (pay-per-call; an OAuth bearer token bypasses payment). Use this for a quick trust read on a Solana agent; for the individual records behind it call solana_agent_attestations, and for identity plus reputation in one call use solana_agent_passport.',
inputSchema: {
type: 'object',
properties: {
asset: { type: 'string', description: 'Metaplex Core asset pubkey (the agent ID)' },
network: { type: 'string', enum: ['mainnet', 'devnet'], default: 'devnet' },
},
required: ['asset'],
additionalProperties: false,
},
async handler(args) {
const data = await solanaReputation(args.asset, args.network || 'devnet');
return {
content: [{ type: 'text', text: JSON.stringify(data, null, 2) }],
structuredContent: data,
};
},
},
{
name: 'solana_agent_attestations',
title: 'List Solana agent attestations',
annotations: ATTESTATION_READ_ANNOTATIONS,
description:
'List recent on-chain attestations about a Solana-registered agent (feedback, validation, task offers, acceptances, disputes). Backed by the three.ws indexer for sub-100ms reads. Each row includes verified/disputed/revoked flags. Public; no auth required. Use this to audit the individual feedback, validation, and dispute records behind a Solana agent reputation score; for the computed summary call solana_agent_reputation.',
inputSchema: {
type: 'object',
properties: {
asset: { type: 'string' },
kind: {
type: 'string',
enum: ['feedback', 'validation', 'task', 'accept', 'revoke', 'dispute', 'all'],
default: 'all',
},
network: { type: 'string', enum: ['mainnet', 'devnet'], default: 'devnet' },
limit: { type: 'integer', minimum: 1, maximum: 200, default: 50 },
},
required: ['asset'],
additionalProperties: false,
},
async handler(args) {
const data = await solanaAttestations({
asset: args.asset,
kind: args.kind || 'all',
network: args.network || 'devnet',
limit: args.limit || 50,
});
return {
content: [{ type: 'text', text: JSON.stringify(data, null, 2) }],
structuredContent: data,
};
},
},
{
name: 'solana_agent_passport',
title: 'Get Solana agent passport',
annotations: ATTESTATION_READ_ANNOTATIONS,
description:
'Full discovery card for a Solana agent: identity (Metaplex Core asset), owner wallet, reputation summary, latest validation result, and attestation schema endpoint. Equivalent to an ERC-8004 passport. Use this when one tool call should answer "who is this agent and can I trust them?".',
inputSchema: {
type: 'object',
properties: {
asset: { type: 'string' },
network: { type: 'string', enum: ['mainnet', 'devnet'], default: 'devnet' },
},
required: ['asset'],
additionalProperties: false,
},
async handler(args) {
const data = await solanaPassport(args.asset, args.network || 'devnet');
return {
content: [{ type: 'text', text: JSON.stringify(data, null, 2) }],
structuredContent: data,
};
},
},
];