Portal Wallet

AdvisoryAudited by Static analysis on May 10, 2026.

Overview

Detected: suspicious.prompt_injection_instructions

Findings (1)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

NoteHigh Confidence

ASI02: Tool Misuse and Exploitation

What this means

If the agent signs or sends the wrong transaction, crypto funds may be lost or transferred irreversibly.

Why it was flagged

The skill documents direct API calls that can send cryptocurrency using the user's MPC share. This is purpose-aligned for a wallet skill, but it is a high-impact action if misused.

Skill content

curl -s -X POST 'https://mpc-client.portalhq.io/v1/assets/send' ... '{share: $share, chain: "monad", to: $to, token: "NATIVE", amount: $amount...}'

Recommendation

Only approve transactions after checking the recipient, chain, asset, amount, and simulation result. Use spending limits and a signature-approval webhook before funding the wallet.

NoteHigh Confidence

ASI03: Identity and Privilege Abuse

What this means

Leaking the API key and a share could let someone else sign transactions from the wallet.

Why it was flagged

The setup guide clearly states that the Portal client API key plus an MPC share grants signing authority. This is expected for the integration but creates a strong credential boundary.

Skill content

anyone with the client API key AND a share can sign transactions

Recommendation

Store these values like wallet private keys, keep openclaw.json out of source control, rotate credentials if exposed, and avoid using production funds until policy controls are configured.

NoteHigh Confidence

ASI08: Cascading Failures

What this means

A bad prompt or unsafe agent behavior could translate into real financial loss if the wallet is funded without containment.

Why it was flagged

The documentation acknowledges that agent compromise or prompt injection could lead to unauthorized wallet actions unless an external policy gate is used.

Skill content

Without a signature approval webhook, this skill should only be used on test wallets with negligible balances. Prompt injection of the agent could otherwise result in unauthorized transactions.

Recommendation

Configure Portal signature approvals, spending caps, allowlisted recipients, and chain restrictions before using non-trivial funds.

Findings (1)

warn

suspicious.prompt_injection_instructions

Location: SKILL.md:26
Finding: Prompt-injection style instruction pattern detected.