Server error 500 on /wallets/sign-typed-data after upgrading legacy agent to ACP v2

[FLX472]

Environment

• SDK: @virtuals-protocol/acp-node-v2 v0.1.3
• @privy-io/node v0.11.0
• viem v2.52.2
• Chain: Base mainnet (chainId 8453)
• Agent upgraded from legacy ACP to v2 on 2026-06-08

What I did

1. Clicked "Upgrade Now" on legacy agent in the EconomyOS dashboard
2. Selected "ACP SDK"
3. Generated key pair — received walletId (23 chars), walletAddress, and signerPrivateKey (PKCS8/PEM format, 184 chars starting with MIGHAgEA)
4. Checked "I have fully configured my agent" and clicked "Confirm Workdone"
5. Configured SDK with all three credentials + builderCode from Identity tab

What happens

Every call to AcpAgent.start() fails immediately with:

BaseError: Server error 500
at serverPost → signedServerCall → signTypedData → SseTransport.authenticate

The authorization signature generates successfully (length 96). Direct curl to the endpoint confirms the 500:

POST https://api.acp.virtuals.io/wallets/sign-typed-data
→ {"statusCode":500,"message":"Internal server error"}

The /auth/agent endpoint responds correctly (returns 400 with validation errors when called without params — as expected).

Question

Is the 23-character alphanumeric walletId the correct format for Privy wallet lookup? Or does the ACP server expect a different identifier? Could there be a propagation delay after confirming the upgrade?

[JessieJanie]

Seeing the exact same failure independently, roughly a month after this was opened — so it does not appear to be resolved, and I do not think it is a walletId-format or propagation issue on your end.

Environment: @virtuals-protocol/acp-node-v2 v0.1.7 (latest), Base mainnet, key-based signer auth. agent.start() fails immediately with BaseError: Server error 500 originating from signTypedData -> SseTransport.authenticate. The authorization signature generates fine locally; /auth/agent and /health both behave correctly.

The important new data point: the entire /wallets/sign-* family returns 500 even with an empty body and no credentials at all, while routing is clearly intact. This rules out walletId format, signer key, and propagation delay — the handlers are crashing server-side before they ever validate input:

curl -i https://api.acp.virtuals.io/health
  -> 200

curl -i -X POST https://api.acp.virtuals.io/wallets/sign-message \
     -H "Content-Type: application/json" -d '{}'
  -> 500 {"statusCode":500,"message":"Internal server error"}

curl -i -X POST https://api.acp.virtuals.io/wallets/sign-typed-data \
     -H "Content-Type: application/json" -d '{}'
  -> 500 {"statusCode":500,"message":"Internal server error"}

curl -i https://api.acp.virtuals.io/wallets/nonexistent
  -> 404   (routing is fine; only the sign-* handlers crash)

Because /auth/agent returns a proper 400 on bad input but the sign-* endpoints 500 on any input, the failure is inside the signing relay itself, not in the client SDK or agent credentials. This blocks agent.start() for anyone on key-based signing.

Could someone from the team confirm the status of the sign-* relay and whether there is a fix/ETA? Happy to provide more detail or test a fix. Thanks.