Back to skill

Security audit

SUPAH Portfolio Guardian

Security checks for vulnerabilities and agentic risk

Overview

This wallet-monitoring skill is not malicious, but it should be reviewed carefully because it overstates its protections and handles wallet-tracking data with weak scoping.

Install only if you are comfortable sending wallet addresses and portfolio queries to SUPAH or a configured endpoint. Do not rely on it for real-time transaction, approval, or suspicious-transfer protection unless the publisher updates the implementation and documentation. Treat the local state file as sensitive wallet-tracking metadata.

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 (4)

T09 · Insecure Skill Coding Practices

Warning
Location
index.js:12
Finding

Unrestricted API Destination Override

Content
View full analysis
{ const options = { hostname: API_BASE, port: 443, path: reqPath, method: 'GET', headers: { 'User-Agent': 'OpenClaw-SUPAH-Portfolio-Guardian/1.2.0' } }; ``` ### Technical Analysis The destination hostname for every portfolio request is read directly from the `SUPAH_API_BASE` environment variable without validation or an allowlist. This conflicts with the skill metadata, which declares outbound access only to `api.supah.ai`. Although HTTPS is used, HTTPS authenticates whichever hostname the application has been configured to contact. It does not prevent an attacker who can influence the process environment from setting the hostname to an attacker-controlled TLS server. Wallet addresses subsequently supplied to portfolio commands would then be transmitted to that server. The unrestricted override also increases the effect of the response-handling weakness because a redirected server can intentionally return malformed, oversized, or indefinitely streaming responses. ### Attack Path 1. An attacker or compromised launcher modifies `SUPAH_API_BASE` in the environment used to start the skill. 2. The value is set to an attacker-controlled hostname with a valid TLS certificate. 3. A user invokes `watch`, `health`, `alerts`, or `rebalance`. 4. `apiRequest()` connects to the attacker-controlled hostname instead of `api.supah.ai`. 5. The attacker receives the requested wallet address and can return manipulated portfolio data or a denial-of-service response. ### Impact Assessment This issue does not directly provide local code execution or access to private wallet keys. Its scope includes: - Disclo ...[truncated 410 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
index.js:57
Finding

Unvalidated Wallet Input Is Embedded in HTTP Request Paths

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
index.js:26
Finding

Unbounded HTTP Response Buffering and Missing Request Timeout

Content
View full analysis
{ const options = { hostname: API_BASE, port: 443, path: reqPath, method: 'GET', headers: { 'User-Agent': 'OpenClaw-SUPAH-Portfolio-Guardian/1.2.0' } }; https.get(options, (res) => { let data = ''; res.on('data', chunk => data += chunk); res.on('end', () => { try { resolve(JSON.parse(data)); } catch (e) { reject(new Error('Invalid JSON response')); } }); }).on('error', reject); }); } ``` ### Technical Analysis The client concatenates all received chunks into a JavaScript string without enforcing a maximum response size. A sufficiently large or endless response can continually consume process memory. No connection or response timeout is configured. A server may accept a connection and then send no data, or send data extremely slowly, leaving the command pending indefinitely. The implementation also attempts to parse every response as JSON without checking the HTTP status code or validating `Content-Type`. Redirects, rate-limit pages, proxy errors, and other non-success responses are therefore handled ambiguously. ### Attack Path 1. The configured endpoint is malicious, compromised, malfunctioning, or redirected through the unrestricted environment override. 2. The endpoint accepts a portfolio request. 3. It returns an extremely large body, continuously streams chunks, or stalls the response. 4. The client continues buffering data or waiting because no size or time limit exists. 5. The process experiences excessive memory consumption, termination, or indefinite unavailability. ### Impact Assessment The primary impact is denial of service within the CLI process: - Unbounded memory growth and ...[truncated 307 chars]
Remediation
View remediation
{ if (res.statusCode < 200 || res.statusCode >= 300) { res.resume(); return reject(new Error(`Unexpected HTTP status: ${res.statusCode}`)); } res.on('data', (chunk) => { received += chunk.length; if (received > MAX_RESPONSE_BYTES) { res.destroy(new Error('Response exceeds size limit')); return; } data += chunk; }); }); request.setTimeout(10_000, () => { request.destroy(new Error('Request timed out')); }); request.on('error', reject); ``` ]]>

T09 · Insecure Skill Coding Practices

Note
Location
index.js:13
Finding

Predictable State File Is Written Without Symlink or Permission Safeguards

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (9)

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The skill advertises continuous real-time wallet monitoring, suspicious transaction detection, approval monitoring, and broad EVM coverage, but the finding indicates the actual behavior is narrower and materially different. In a crypto security context, this can cause users to rely on protections that are not actually present, potentially missing risky approvals, malicious transfers, or chain-specific threats while believing they are being monitored.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The manifest promises detection of suspicious transactions and monitoring of approvals, which implies transaction- or allowance-level analysis. The code never inspects transaction history, approval events, spender allowances, or similar data; it only reads portfolio summaries, token scores, and derives alerts from changes in aggregate risk scores.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
85% confidence
Finding

The skill declares environment-variable and external-network capabilities in metadata but does not define an explicit tool scope such as permissions or allowed-tools. That weakens least-privilege boundaries and makes it harder for users or hosting agents to understand what the skill is permitted to access, which is especially important for a crypto-related skill that transmits wallet data to a third-party API.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill describes privacy in terms of not storing private keys, but it does not clearly warn that wallet addresses, balances, transaction history, approvals, and portfolio activity may be sent to external services such as SUPAH and Moralis. Wallet addresses are highly sensitive in a crypto context because they reveal holdings and behavior, enabling profiling, doxxing correlations, and targeted scams.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The usage examples use broad natural-language trigger phrases without clear activation constraints, which can cause accidental invocation or overbroad matching by an agent. Because this skill can send wallet addresses and portfolio activity to an external paid API, unintended activation could leak sensitive financial metadata and trigger unwanted micropayments.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The module documentation asserts that access is mediated by x402 USDC micropayments on Base, implying payment-enforced API usage. In practice, apiRequest sends a plain HTTPS GET with only a User-Agent header and no payment token, x402 handling, or related authentication/payment logic.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill writes watched wallet addresses, timestamps, thresholds, and last risk scores to a predictable plaintext file in the user's home directory without notice or permission controls. In a wallet-monitoring context, this creates a privacy leak because local users, backups, logs, or other software on the host may infer which wallets are being tracked and associated risk history.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

Wallet addresses and derived portfolio queries are sent to a remote API endpoint without explicit disclosure at the call sites or visible consent handling, and the destination host can be changed via an environment variable. In the context of a portfolio guardian, this is sensitive behavioral and financial metadata; silent transmission to third parties can expose user holdings and monitoring activity, and the configurable host increases the chance of accidental or malicious data exfiltration.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The manifest describes real-time monitoring and alerts across Base and EVM chains, but the implementation only performs manual checks when the user invokes commands like watch, alerts, or health. It persists watched wallets locally and queries a single /agent/v1/portfolio/<address> API endpoint, with no scheduler, push/notification mechanism, or multi-chain handling visible in code.

Content

No source excerpt is available for this finding.