Back to skill

Security audit

AI Trading

Security checks for vulnerabilities and agentic risk

Overview

This skill stays within human-reviewed stock research, but verified calculation errors mean its figures and position sizes need independent checking.

Use market-data keys only. Independently verify volume, historical ranges, and position sizes; the overnight flag does not halve shares, and the stated risk rules are not all enforced. Live API support also needs repair. These verified defects warrant caution with the research output, but the inspected bundle shows no hidden execution, credential theft, or persistence.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (6)

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · references/data-sources.md (reported line 19)May include surrounding context.

bash
curl -H "User-Agent: your-email@example.com" \
  "https://data.sec.gov/api/xbrl/companyconcept/CIK0000320193/us-gaap/Revenues.json"
  • company_tickers.json maps a ticker to a CIK.

Missing volume data is converted to zero and included in reported volume averages.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

An empty CSV volume field is silently converted to zero, and that fabricated observation is included in the reported average volume. In a financial-analysis skill, this can understate trading activity and mislead liquidity assessments; similar missing-to-zero conversions also occur in provider parsing.

Content

No source excerpt is available for this finding.

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 · scripts/market_snapshot.py (reported line 190)May include surrounding context.

python
def fetch_twelvedata(symbol: str, key: str) -> tuple:
    quote = http_json(
        f"https://api.twelvedata.com/quote?symbol=%7Burllib.parse.quote%28symbol%29%7D&apikey=%2A%2A%2A}"
    )
    series = http_json(
        f"https://api.twelvedata.com/time_series?symbol=%7Burllib.parse.quote%28symbol)}"

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 · scripts/market_snapshot.py (reported line 193)May include surrounding context.

python
f"https://api.twelvedata.com/quote?symbol=%7Burllib.parse.quote%28symbol%29%7D&apikey=%2A%2A%2A}"
    )
    series = http_json(
        f"https://api.twelvedata.com/time_series?symbol=%7Burllib.parse.quote%28symbol)}"
        f"&interval=1day&outputsize=400&apikey={key}"
    )
    bars = []

Reports arbitrary-history extrema as 52-week metrics despite promising unavailable values will not be estimated.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The snapshot labels extrema from all supplied bars as 52-week highs and lows without checking their dates or ensuring sufficient historical coverage. Short or excessively long CSV histories therefore produce misleading financial metrics, including the percentage from the 52-week high, that downstream users or agents could mistake for validated market indicators.

Content

No source excerpt is available for this finding.

The --overnight help promises to halve the recommended size, but the generated plan retains the full share count.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The --overnight option promises to halve the recommended size, but build_plan leaves max_shares and all associated exposure calculations unchanged; it only adds a warning when risk_pct exceeds 0.5. In this financial-risk tool, users or downstream consumers relying on the numeric plan could therefore take twice the promised overnight position size, although the manual-approval requirement and absence of order execution limit direct impact.

Content

No source excerpt is available for this finding.