Back to skill

Security audit

SUPAH Whale Tracker

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent paid crypto analytics tool, but it can make automatic USDC micropayments and send wallet or token queries to a configurable API host without enough user control or destination scoping.

Install only if you are comfortable with automatic x402 USDC payments for API calls. Set a spending cap or require confirmation in your agent if possible, avoid setting SUPAH_API_BASE unless you trust the destination, and treat wallet or token queries as information that may be sent to the configured API provider.

Vulnerability Patterns
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (2)

T09 · Insecure Skill Coding Practices

Warning
Location
index.js:7
Finding

Unrestricted API Base URL Allows Requests to Attacker-Controlled Hosts

Content
View full analysis
{ const qs = new URLSearchParams(params).toString(); const url = `${API}${path}${qs ? '?' + qs : ''}`; https.get(url, { headers: { 'User-Agent': 'OpenClaw-SUPAH-Whale-Tracker/1.2.0' } }, res => { ``` ### Technical Analysis The `SUPAH_API_BASE` environment variable is accepted without validating its protocol, hostname, port, credentials, or path components. It is then concatenated directly with an API path and used as the destination of an outbound HTTPS request. This behavior conflicts with the declared outbound network boundary in `SKILL.md`, which identifies only `api.supah.ai`. An attacker or compromised deployment configuration capable of setting `SUPAH_API_BASE` can redirect requests to an arbitrary HTTPS server. User-supplied token identifiers and wallet addresses are incorporated into endpoint paths for the `track`, `watch`, and `analyze` commands. Redirecting the base URL can therefore disclose those queries to an unauthorized service. The hostile service can also return fabricated JSON that the CLI prints without establishing that it came from the intended SUPAH API. ### Attack Path 1. An attacker gains control over the environment used to launch the skill, or supplies a malicious deployment configuration. 2. The attacker sets `SUPAH_API_BASE` to an attacker-controlled HTTPS origin. 3. A user invokes a command such as `watch 0x...`, `track TOKEN`, or `analyze TOKEN`. 4. The skill appends the command-derived endpoint to the malicious base URL. 5. The wallet address, token identifier, and request intent are transmitted to the attacker-controlled server. 6. The server returns arbitrar ...[truncated 758 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
index.js:12
Finding

Unbounded and Untimed HTTP Response Processing Enables Denial of Service

Content
View full analysis
{ let d = ''; res.on('data', c => d += c); res.on('end', () => { try { resolve(JSON.parse(d)); } catch (e) { resolve({ error: 'Invalid response' }); } }); }).on('error', reject); ``` ### Technical Analysis The HTTP client accumulates the entire response body in the `d` string without enforcing a maximum size. It also does not configure a connection or response timeout. As a result: - A server can continuously stream data and cause memory consumption to grow until the Node.js process becomes unstable or terminates. - A server can accept the connection but indefinitely delay or never complete the response, leaving the CLI invocation stalled. - The code does not validate HTTP status codes or response content types before buffering and parsing the body. The risk is increased by the configurable `SUPAH_API_BASE`, which can direct the client to an attacker-controlled HTTPS service. The flaw can also be triggered if the legitimate API or an upstream component is compromised or malfunctioning. ### Attack Path 1. The client sends a request to a malicious, compromised, or malfunctioning API endpoint. 2. The endpoint either keeps the response open without completing it or continuously sends response data. 3. Because no timeout is configured, the process can remain blocked indefinitely. 4. Because no maximum response size is enforced, every received chunk is appended to an in-memory string. 5. Memory consumption grows until the request is interrupted, the process crashes, or the host experiences resource pressure. ### Impact Assessment The vulnerability affects availability rather than confidentiality or local privilege boundaries. A successful exploit can: - Hang an ind ...[truncated 442 chars]
Remediation
View remediation
{ if (res.statusCode < 200 || res.statusCode >= 300) { res.resume(); reject(new Error(`Unexpected HTTP status: ${res.statusCode}`)); return; } const contentType = res.headers['content-type'] || ''; if (!contentType.includes('application/json')) { res.resume(); reject(new Error('Unexpected response content type')); return; } let size = 0; let data = ''; res.on('data', chunk => { size += chunk.length; if (size > MAX_RESPONSE_BYTES) { res.destroy(new Error('Response exceeds size limit')); return; } data += chunk; }); res.on('end', () => { try { resolve(JSON.parse(data)); } catch { reject(new Error('Invalid JSON response')); } }); res.on('error', reject); }); req.on('timeout', () => { req.destroy(new Error('Request timed out')); }); req.on('error', reject); ``` ]]>
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (3)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
87% confidence
Finding

The skill declares environment-variable and binary requirements plus outbound network access, but does not define an explicit tool/permission scope. That creates ambiguity about what the agent may access at runtime and can lead to over-broad execution privileges, especially in hosts that infer capabilities from metadata rather than enforcing least privilege. In a paid, networked skill, unclear scope increases the chance of unintended external calls or access to sensitive environment data.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The invocation examples are broad natural-language prompts like 'Show me recent whale moves on Base' and 'Track this wallet', with no trigger boundaries, confirmation requirements, or exclusions. In an agent environment, this makes accidental invocation more likely and can cause the skill to fire on loosely related user requests, leading to unintended network activity and paid API calls. Because the skill is connected to x402 micropayments, overly broad activation is more dangerous than a normal read-only skill.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The pricing section mentions x402 micropayments, but the skill description and usage flow do not clearly warn that invoking the skill causes automatic paid transactions. Users may issue ordinary analytics questions without realizing each call can spend funds, which creates a real risk of surprise charges or repeated small losses through accidental or excessive invocation. The combination of natural-language triggering and automatic payment materially increases the risk.

Content

No source excerpt is available for this finding.