Back to skill

Security audit

Gate Exchange Dual Investment

Security checks for vulnerabilities and agentic risk

Overview

This Gate dual-investment skill is a disclosed financial workflow that can place orders only through a scoped MCP tool after explicit user confirmation.

Install only if you intend to use Gate dual investment with a locally configured Gate MCP session and Earn:Write API permission. Review any order draft carefully because confirmed dual-investment orders commit funds, cannot be canceled after placement, and may settle in a different currency.

Vulnerability Patterns
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (12)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 74)May include surrounding context.

md
- `SKILL.md` keeps routing and product semantics.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
85% confidence
Finding

The trigger list is broad enough that ordinary questions about target prices, buy-low/sell-high strategies, or dual balances could invoke a skill with write-capable financial tooling. In this context, unintended invocation increases the chance that the agent enters a transactional workflow or surfaces account-linked data when the user only wanted general information.

Content

No source excerpt is available for this finding.

Unrestricted Tool Access

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

Skill grants unrestricted tool access without appropriate constraints. An agent with unfettered tool access can perform arbitrary actions including file modification, network requests, and code execution.

Content

Scanner excerpt · SKILL.md (reported line 29)May include surrounding context.

md
## General Rules

⚠️ STOP — You MUST read and strictly follow the shared runtime rules before proceeding.
Do NOT select or call any tool until all rules are read. These rules have the highest priority.
→ Read `./references/gate-runtime-rules.md`
- **Only call MCP tools explicitly listed in this skill.** Tools not documented here must NOT be called, even if they
  exist in the MCP server.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
80% confidence
Finding

The markdown documents a write-capable order-placement tool, but the user-facing warning about irreversibility appears later in the domain knowledge section rather than as a prominent warning tied to execution behavior. Because order placement affects user funds and cannot be canceled, the description should clearly foreground this risk in its operational warnings.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 59)May include surrounding context.

md
- Credentials Source: Local Gate MCP deployment (`GATE_API_KEY`, `GATE_API_SECRET`)
- API Key Required: Yes
- Permissions: Earn:Write
- Never ask the user to paste secrets into chat; rely on the configured MCP session only.
- API Key Provisioning Reference: https://www.gate.com/myaccount/profile/api-key/manage (create or rotate keys outside the chat when the local MCP setup requires them).

### Installation Check

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 160)May include surrounding context.

md
- **APY sanity check (MANDATORY before responding)**: After formatting ALL APY values, scan every value in your output. Typical correct ranges after ×100: crypto sell-high plans → 10%–2000%; stablecoin buy-low plans → 5%–1800%. **If you see any APY displayed as 0.05%–20% (single or low-double digits), you almost certainly forgot to multiply by 100. STOP, go back, and recompute ALL APY values before responding.** For example, if a raw value is `19.9378`, the correct display is `1993.78%` — NOT `19.94%`.
- **No investment advice**: Do not recommend specific plans or predict prices. Present data and let the user decide.
- **Non-principal-protected**: Always clearly communicate that dual investment is NOT principal-protected and the user may receive a different currency.
- **Order placement confirmation**: Before calling `cex_earn_place_dual_order`, MUST show the user the full order details (plan, amount, target price, APY, settlement scenarios) and get **explicit user confirmation**. NEVER place an order without confirmation.
- **Sensitive data**: Never expose API keys, internal endpoint URLs, or raw error traces to the user.

## Error Handling

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · references/gate-runtime-rules.md (reported line 16)May include surrounding context.

md
## 1. MCP Session and Authentication

- Use the already configured Gate MCP session for the current host.
- Local Gate MCP deployments use `GATE_API_KEY` and `GATE_API_SECRET`; never ask the user to paste these secrets into chat.
- Minimal permission for this skill is `Earn:Write`.
- If the required Gate MCP tools are missing, stop write actions and switch to setup guidance only.
- If the MCP session returns an auth or permission error, stop write actions and guide the user to repair the configured local MCP credentials before continuing.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The file explicitly states for cex_earn_list_dual_orders that there is no type field and that sell-high/buy-low must be derived from invest_currency (L045-L049). In Case 3, the instructions then tell the agent to filter by type=call or type=put, which contradicts the documented data model and could lead to behavior based on a nonexistent field.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
99% confidence
Finding

Earlier, the document clearly says there is no instrument_name field and that coin matching must use invest_currency or exercise_currency instead (L049). The reporting templates for product listing and settlement simulation nevertheless include {instrument_name}, directly contradicting the stated available fields.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The browse/list trigger phrases are broad and can match ordinary informational requests about dual products without strong boundaries, increasing the chance this skill is invoked when the user did not intend a transactional or account-linked workflow. In a financial-product skill, unintended invocation can expose confusing investment guidance, cause incorrect tool selection, or prime the agent toward risky product flows the user did not explicitly request.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

These order-placement examples are open-ended enough that a casual statement of trading intent could activate an order workflow for a complex financial product. Because the skill can ultimately place orders after a confirmation step, overbroad triggering materially raises the risk of the agent entering a transactional flow the user did not clearly request.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The buy-low examples overlap with general trading discussion, especially for users describing a market view rather than requesting a dual-investment product. In this context, ambiguity is more dangerous because the skill concerns leveraged-like settlement outcomes and can lead the agent into a product subscription workflow for a regulated financial instrument.

Content

No source excerpt is available for this finding.