Back to skill

Security audit

APEX IA Pro

Security checks for vulnerabilities and agentic risk

Overview

This skill advertises a Binance Futures scanner, but the package also contains under-disclosed automated trading programs with embedded Binance credentials and default-on leveraged execution paths.

Review this as more than a scanner. Do not install or run the trader scripts unless you intentionally want automated Binance Futures trading, have rotated any exposed credentials, understand whether testnet or live endpoints are configured, and have explicit controls for order confirmation, leverage, stop loss, and account permissions.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (6)

T09 · Insecure Skill Coding Practices

Error
Location
apex-ia-trader.mjs:14
Finding

Plaintext Binance API credentials embedded in multiple executable files

Content
View full analysis
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Error
Location
apex-ia-aggressive.mjs:21
Finding

Default-enabled automatic leveraged trading exceeds the declared scanner scope

Content
View full analysis
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Error
Location
start-apex-completo.sh:6
Finding

Complete launcher starts an account-authorized trader as a background process

Content
View full analysis
bridge-signals.json # Iniciar o trader em background node apex-ia-trader.mjs & TRADER_PID=$! # Aguardar 2 segundos sleep 2 # Iniciar o scanner node apex-ia-software.mjs # Quando o scanner fechar, matar o trader kill $TRADER_PID 2>/dev/null ``` The trader initializes its mode as automatic: ```js let autoMode = true; setInterval(async () => { if (autoMode) { await getBalance(); await monitorPositions(); drawTraderUI(); } }, 2000); ``` ### Technical Analysis The “complete” launcher starts `apex-ia-trader.mjs` in the background before starting the scanner. The background process immediately uses authenticated Binance functionality to obtain account data and monitor its in-memory positions. The reviewed `apex-ia-trader.mjs` exports order-execution functions, but the inspected launcher and scanner do not demonstrate a direct call from `apex-ia-software.mjs` into those functions. Therefore, the confirmed automatic effect of this launcher is background startup and account access; automatic order placement through this particular launcher requires an additional signal-integration call not established in the reviewed path. Even with that limitation, launching an account-authorized trading process as a side effect of starting a scanner is inconsistent with least privilege and creates a dangerous operational boundary. ### Attack Path 1. A user runs `start-apex-completo.sh` to start the advertised complete scanner system. 2. The script starts `apex-ia-trader.mjs` as a background process. 3. The ...[truncated 1035 chars]
Remediation
View remediation

T08 · Insecure Dependencies

Warning
Location
realtime-scanner.js:154
Finding

Runtime npm installation executes mutable third-party dependency code

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
webapp/server.js:14
Finding

Unauthenticated web API permits cross-origin signal access and outbound request abuse

Content
View full analysis
{ const { symbol = 'BTCUSDT', interval = '1h', limit = 50 } = req.query; try { const response = await axios.get(`https://api.binance.com/api/v3/klines`, { params: { symbol, interval, limit } }); const klines = response.data.map(k => ({ time: k[0], open: parseFloat(k[1]), high: parseFloat(k[2]), low: parseFloat(k[3]), close: parseFloat(k[4]), volume: parseFloat(k[5]) })); res.json(klines); } catch (err) { res.status(500).json({ error: err.message }); } }); ``` The `/api/signals` route exposes locally stored signal history: ```js app.get('/api/signals', (req, res) => { try { if (fs.existsSync(SIGNALS_FILE)) { const signals = JSON.parse(fs.readFileSync(SIGNALS_FILE, 'utf-8')); res.json(signals.slice(-50).reverse()); } else { res.json([]); } } catch (err) { res.status(500).json({ error: err.message }); } }); ``` ### Technical Analysis The server has no authentication or rate limiting. `cors()` permits browser scripts from any origin to read API responses when they can reach the service. The kline route accepts arbitrary `symbol`, `interval`, and `limit` values and forwards them to Binance without type checks, allowlists, or application-level bounds. The outbound host is fixed to Binance, so this is not general server-side request forgery. It can nevertheless be abused as a request r ...[truncated 1333 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
src/scanner.ts:327
Finding

Unbounded Agent-controlled scan parameters can amplify Binance API requests

Content
View full analysis
{ const signals = await scanAll(args); ``` The values are used directly: ```ts export async function scanAll(options: ScanOptions = {}): Promise { const { minScore = 6, includeTFs = ['15m', '1h', '4h'], symbolLimit = 30 } = options; console.log(`🔍 Escaneando até ${symbolLimit} pares...`); let symbols = await fetchSymbols(); if (symbolLimit && symbols.length > symbolLimit) { symbols = symbols.slice(0, symbolLimit); } const allSignals: FinalSignal[] = []; const batchSize = 3; for (let i = 0; i < symbols.length; i += batchSize) { const batch = symbols.slice(i, i + batchSize); const batchResults = await Promise.all( batch.map(symbol => scanSymbol(symbol, includeTFs)) ); allSignals.push(...batchResults.flat()); } ``` For every supplied timeframe value, `scanSymbol` makes another request: ```ts for (const tf of timeframes) { try { const candles = await fetchKlines(symbol, intervalMap[tf] || '1h', 100); const signal = calculateFinalSignal(symbol, tf, candles); if (signal && signal.score >= 6) { results.push(signal); } await new Promise(resolve => setTimeout(resolve, 100)); } catch (error) { // Silencia erros individuais } } ``` ### Technical Analysis TypeScript interfaces are comp ...[truncated 1528 chars]
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (200)

Description-Behavior Mismatch

Critical
Category
Not specified by scanner
Confidence
99% confidence
Finding

The skill is described as a market scanner, but the code enables automated leveraged futures trading by default via autoTrade = true and related position-management logic. This is dangerous because users expecting passive analysis could unknowingly authorize live or testnet account actions, and the mismatch obscures the true risk profile of the skill.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Critical
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill contains signed Binance request support and account-changing API operations inconsistent with a scanner-only description. This is dangerous because it gives the skill the ability to modify leverage, place orders, and affect account state while appearing to be a read-only analysis tool.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Critical
Category
Not specified by scanner
Confidence
99% confidence
Finding

These functions open and close market positions, calculate targets, and maintain internal position state, clearly exceeding the described scope of scanning. In context, this is especially dangerous because futures market orders with leverage can create immediate financial exposure and losses without users realizing the skill is an execution bot.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The implemented behavior appears narrower and different than advertised: fixed 1h timeframe only, spot API instead of Futures, only SMA crossover logic, undeclared local signal file reads, and extra API endpoints beyond the described scanner. This mismatch is dangerous because it undermines informed consent, can expose local data through filesystem access, and may lead users to rely on incorrect market assumptions.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The implemented behavior appears narrower and different than advertised: fixed 1h timeframe only, spot API instead of Futures, only SMA crossover logic, undeclared local signal file reads, and extra API endpoints beyond the described scanner. This mismatch is dangerous because it undermines informed consent, can expose local data through filesystem access, and may lead users to rely on incorrect market assumptions.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The implemented behavior appears narrower and different than advertised: fixed 1h timeframe only, spot API instead of Futures, only SMA crossover logic, undeclared local signal file reads, and extra API endpoints beyond the described scanner. This mismatch is dangerous because it undermines informed consent, can expose local data through filesystem access, and may lead users to rely on incorrect market assumptions.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The implemented behavior appears narrower and different than advertised: fixed 1h timeframe only, spot API instead of Futures, only SMA crossover logic, undeclared local signal file reads, and extra API endpoints beyond the described scanner. This mismatch is dangerous because it undermines informed consent, can expose local data through filesystem access, and may lead users to rely on incorrect market assumptions.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The implemented behavior appears narrower and different than advertised: fixed 1h timeframe only, spot API instead of Futures, only SMA crossover logic, undeclared local signal file reads, and extra API endpoints beyond the described scanner. This mismatch is dangerous because it undermines informed consent, can expose local data through filesystem access, and may lead users to rely on incorrect market assumptions.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The implemented behavior appears narrower and different than advertised: fixed 1h timeframe only, spot API instead of Futures, only SMA crossover logic, undeclared local signal file reads, and extra API endpoints beyond the described scanner. This mismatch is dangerous because it undermines informed consent, can expose local data through filesystem access, and may lead users to rely on incorrect market assumptions.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
100% confidence
Finding

Hardcoded Binance API credentials are embedded directly in the skill file, making them trivially extractable by anyone with file access, logs, backups, or repository history. In a trading context this is especially dangerous because the credentials authorize authenticated exchange actions, potentially including account access and order placement.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

Using hardcoded exchange credentials for authenticated requests without clear user disclosure creates both a secret-exposure risk and an unauthorized-action risk. Anyone reusing or obtaining the file may unknowingly operate with embedded credentials or exfiltrate them for direct account abuse.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill is presented as a market scanner, but this code also places leveraged futures orders, manages positions, and closes trades automatically. That mismatch is dangerous because users may grant trust or run the skill expecting passive analysis, while it can perform destructive financial actions with real exchange APIs if demo mode is changed or reused elsewhere.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill enables automatic leveraged trading without per-trade confirmation, while configured for aggressive behavior and with stop loss disabled by default. In this context, silent execution is a serious safety issue because market orders can create immediate financial loss before the user realizes the tool is acting autonomously.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The code hardcodes an API key and performs signed exchange requests without any prominent warning, consent flow, or safety disclosure. This is dangerous because users may run the skill without realizing it can authenticate to exchange endpoints and potentially act on account data or trading permissions.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The skill is described as a scanner, but this code contains full trade execution and position-closing logic for Binance Futures orders. That mismatch is dangerous because users may grant trust or credentials expecting passive market analysis while the code is capable of taking direct trading actions on their account.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The auto-trading path expands the skill from signal generation into autonomous execution of futures trades. In the context of a tool presented as a scanner, this creates a harmful capability gap where financial loss can occur without users understanding they enabled order placement.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

Automatic trade execution can place market orders without a per-order confirmation prompt, creating a direct path to unintended financial loss. This is particularly dangerous for futures trading, where leverage and rapid price movement can amplify losses beyond what users expect from a scanning tool.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The main loop automatically executes trades from generated signals when autoTrade is enabled, turning scan results into live actions. This is especially risky because the overall skill description suggests an analytical scanner, not an autonomous trading bot operating on a derivatives exchange.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
100% confidence
Finding

Hardcoded Binance API credentials are embedded directly in source code. Even if USE_DEMO is currently true, exposed credentials can be copied, reused, or accidentally promoted to production patterns later, and they normalize unsafe secret handling in code distributed to others.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The code uses hardcoded API credentials without any warning, consent prompt, or disclosure to the user. This is dangerous because users may run the skill unaware that private exchange credentials are embedded and actively used for authenticated requests.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

Automatic trading is enabled by default, and the main loop will execute market orders based on generated signals without explicit confirmation. This is dangerous because default-on auto-execution can cause unintended leveraged trades immediately after startup.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest claims confirmation using Pivot SuperTrend, RSI, volume, and confluence, but the implemented scan logic only evaluates SMA 8/21 cross/touch conditions. This discrepancy is dangerous because it misrepresents how signals are produced, causing users to trust trades based on controls that do not actually exist.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
100% confidence
Finding

Hardcoded API credentials in source code are a direct secret-exposure vulnerability. Anyone with access to the file can reuse the credentials to access the linked Binance account on the configured environment, potentially reading balances and placing or managing futures orders.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The code includes signed account-access and trading-request capability via API key/secret handling even though the skill is presented as a scanner. In this context, unnecessary authenticated exchange access greatly increases risk because the skill can query private account data and submit trades beyond the user's likely expectations.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill is described as a scanner, but it contains active trade execution and position-closing functions that submit authenticated Binance Futures market orders. This mismatch is dangerous because users may grant trust or run the skill expecting passive analysis while it can perform irreversible financial actions on their account.

Content

No source excerpt is available for this finding.