Back to skill

Security audit

NEXUS Ap2 Batched Settle

Security checks for vulnerabilities and agentic risk

Overview

This payment skill is coherent but needs review because it asks users to trust a remote service's unsigned wallet transactions for batched payments without clear pre-sign validation or sandbox/mainnet separation.

Install only if you trust the NEXUS service and will manually inspect every wallet signing request. Before signing, verify the decoded transaction fields yourself: destination, amount, currency, issuer, fee, network, memo, and every inner payment in a batch. Do not use sandbox_ signatures or automated signing for production funds unless the service provides a clearly isolated sandbox/mainnet boundary and wallet-side limits.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Findings (2)

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:118
Finding

Prefix-Based Sandbox Signature Bypass Exposed Through the Payment Mandate API

Content
View full analysis
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Error
Location
SKILL.md:145
Finding

Unsigned Financial Transactions Are Trusted From an External Construction Service Without Required Field-Level Validation

Content
View full analysis
", "recipient_address": "rM86ChiqozvfHwLckKRq2yTtzKLwfBe2XA", "chain": "xrpl" }' ``` The returned object is then handled as follows: ```markdown Returns `unsigned_tx` — an XRPL Payment JSON with hex-encoded `ap2.mandate.id` memo. Sign with your wallet (Xumm, Crossmark, hardware), broadcast, then confirm: ``` `SKILL.md:165-178`: ```bash curl -X POST https://ai-service-hub-15.emergent.host/api/ap2/payments/batch \ -H "Content-Type: application/json" \ -d '{ "mandate_ids": ["ap2_cart_001", "ap2_cart_002", "ap2_cart_003", "ap2_intent_001"], "payer_address": "r", "chain": "xrpl", "mode": "ALLORNOTHING" }' ``` ```markdown Returns an unsigned **XLS-56 Batch** transaction containing all inner Payments. Sign + broadcast once. ``` `SKILL.md:194-198`: ```bash curl -X POST https://ai-service-hub-15.emergent.host/api/xrpl/build/trust-line \ -H "Content-Type: application/json" \ -d '{"holder_address":"r","currency":"RLUSD"}' ``` ```markdown Then use `"currency": "RLUSD"` in your mandates. RLUSD issuer is `rMxCKbEDwqr76QuheSUMdEGf4B9xJ8m5De`. ``` ### Tec ...[truncated 2802 chars]
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (6)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The README promotes batched settlement of up to 20 payments in one signed transaction but does not prominently warn that this can move real wallet funds, may be irreversible once signed and submitted, and can amplify loss by bundling multiple payments atomically. In a payments skill, missing consent and risk disclosure increases the chance that users or agents authorize financially significant actions without understanding the consequences.

Content

No source excerpt is available for this finding.

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · README.md (reported line 28)May include surrounding context.

Or manually:

bash
cp SKILL.md ~/.openclaw/skills/nexus-ap2-batched-settle/SKILL.md

Quick start

External Transmission

Medium
Category
Data Exfiltration
Confidence
86% confidence
Finding

The quick start instructs users to POST mandate creation data, including identifiers and signature material, to an external service endpoint. Even though the example appears to use sandbox/demo values, it normalizes transmitting payment authorization metadata to a third-party host without privacy, trust-boundary, retention, or validation warnings.

Content

Scanner excerpt · README.md (reported line 35)May include surrounding context.

bash
# 1. Mint a cart mandate
MID=$(curl -sX POST https://ai-service-hub-15.emergent.host/api/ap2/mandates/create \
  -H "Content-Type: application/json" \
  -d '{
    "mandate_type": "cart",

External Transmission

Medium
Category
Data Exfiltration
Confidence
94% confidence
Finding

This example sends mandate IDs and the payer wallet address to a remote API to build a batch payment transaction, directly involving payment flow and wallet-linked metadata. In the context of an atomic settlement skill, this is more dangerous because a compromised or untrusted service could influence transaction construction, misroute funds, or facilitate unintended settlement preparation at scale.

Content

Scanner excerpt · README.md (reported line 46)May include surrounding context.

}' | jq -r .mandate_id)

2. Build unsigned batch tx (using multiple mandate IDs)

curl -X POST https://ai-service-hub-15.emergent.host/api/ap2/payments/batch
-H "Content-Type: application/json"
-d "{"mandate_ids":["$MID"],"payer_address":"r","chain":"xrpl","mode":"ALLORNOTHING"}"

text

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 99)May include surrounding context.

For an Intent Mandate (agent autonomy, e.g. "spend up to 10 XRP this week"):

bash
curl -X POST https://ai-service-hub-15.emergent.host/api/ap2/mandates/create \
  -H "Content-Type: application/json" \
  -d '{
    "mandate_type": "intent",

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The RLUSD setup instructions tell the user to create a trust line but do not clearly warn that this is a state-changing on-ledger action that can consume wallet reserve and alter account exposure. In a payment skill, omission of that warning can mislead users into signing a transaction they may not understand, especially because trust lines are persistent wallet configuration changes rather than simple read-only setup.

Content

No source excerpt is available for this finding.