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
docs(wallet): document the sweeper-bot withdrawal warning and the base58 secret-scanner rule
nirholas9d3234104195
diff --git a/docs/agent-abilities/wallet/18-withdraw.md b/docs/agent-abilities/wallet/18-withdraw.mdindex 6d6cc04a5..787ce8023 100644--- a/docs/agent-abilities/wallet/18-withdraw.md+++ b/docs/agent-abilities/wallet/18-withdraw.md@@ -24,6 +24,7 @@ The agent's private key lives encrypted on the server and never touches the brow- Client-side validation: positive amounts only, amount cannot exceed the available balance, destination must resolve- Two-step flow: a review/confirm screen summarizing asset, amount, destination, and network before anything is submitted- Live allowlist badge on the confirm screen: '✓ allowlisted' or '⚠ not on allowlist — this will be rejected' before you even press Confirm+- Sweeper-bot warning on the confirm screen: a read-only check of the destination's recent on-chain history flags a wallet that is repeatedly drained within seconds of receiving funds, with a blocking confirm you must explicitly accept before the send goes through- Irreversibility warning on every confirmation: crypto transfers are final- Risk-acknowledgment gate before any mainnet withdrawal (with a graceful fallback prompt if the dialog module fails to load); devnet skips it- Single-use CSRF token on every state-changing request — the server burns each token on use@@ -51,7 +52,7 @@ The agent's private key lives encrypted on the server and never touches the brow## Guardrails & safety-Owner-only end to end: every endpoint verifies the signed-in user owns the agent. Withdrawals are capped at 5 per day per user with an additional per-IP burst guard. Every state-changing request requires a single-use CSRF token, and mainnet withdrawals require an explicit risk acknowledgment. The server enforces the shared spend policy before signing: withdraw allowlist (if set, funds can only go to approved addresses), per-transaction and rolling 24-hour USD ceilings, the owner's plain-English safety rules enforced deterministically, and a behavioral anomaly guard that can auto-freeze the wallet. The freeze switch halts all autonomous spending but deliberately never blocks the owner's withdrawals — a freeze can never trap funds. Destinations must be valid, on-curve addresses and cannot be the wallet itself. SOL sweeps always reserve rent plus fees so the account survives. Idempotency keys make retries safe against double-sends, ambiguous confirmations return a pending state with an explorer link instead of guessing, and every withdrawal, key decryption, limit change, and policy block lands in an owner-visible audit ledger.+Owner-only end to end: every endpoint verifies the signed-in user owns the agent. Withdrawals are capped at 5 per day per user with an additional per-IP burst guard. Every state-changing request requires a single-use CSRF token, and mainnet withdrawals require an explicit risk acknowledgment. The server enforces the shared spend policy before signing: withdraw allowlist (if set, funds can only go to approved addresses), per-transaction and rolling 24-hour USD ceilings, the owner's plain-English safety rules enforced deterministically, and a behavioral anomaly guard that can auto-freeze the wallet. The freeze switch halts all autonomous spending but deliberately never blocks the owner's withdrawals: a freeze can never trap funds. Destinations must be valid, on-curve addresses and cannot be the wallet itself. The server also reads the destination's recent transaction history for the sweeper-bot pattern (incoming SOL drained again within seconds, repeated); a flagged address blocks the send with an explicit confirm the owner must accept before it is retried. SOL sweeps always reserve rent plus fees so the account survives. Idempotency keys make retries safe against double-sends, ambiguous confirmations return a pending state with an explorer link instead of guessing, and every withdrawal, key decryption, limit change, and policy block lands in an owner-visible audit ledger.## Screenshot-worthy (shot list)diff --git a/docs/ops/hosting-and-credentials.md b/docs/ops/hosting-and-credentials.mdindex 062f24272..4c00126b2 100644--- a/docs/ops/hosting-and-credentials.md+++ b/docs/ops/hosting-and-credentials.md@@ -125,9 +125,14 @@ token, create a scoped replacement with the narrowest required package permissioaccount's token activity.The pre-push guard at `scripts/check-secrets.mjs` rejects provider token formats and R2 credential-pairs in added content. The runtime `redactUrlSecrets` guard also masks those formats when they-arrive inside free-form text destined for structured logs. Both are defense in depth, not a-substitute for rotation after disclosure.+pairs in added content. It also catches a base58-encoded Solana keypair secret key: any 86-90+character base58 run is decoded, and flagged only when the first 32 bytes derive the embedded+public key in the last 32 bytes via Ed25519, so the thousands of real 64-byte transaction+signatures already in the repo never trigger a false positive. The runtime `redactUrlSecrets`+guard also masks those formats when they arrive inside free-form text destined for structured+logs. Both are defense in depth, not a substitute for rotation after disclosure. A caught Solana+secret follows the same incident response above: treat the exposed keypair as compromised, move+its funds to a freshly generated wallet, and never reuse the key.## Production verification after any credential change