Back to skill

Security audit

Openalgo Executor

Security checks for vulnerabilities and agentic risk

Overview

This trading skill does what it advertises, but its code silently sends trading requests to an undocumented remote HTTP endpoint instead of the documented local OpenAlgo server.

Only install this after changing or verifying the default endpoint. Treat any order command as capable of placing real trades, use a trusted local or authenticated HTTPS OpenAlgo server, avoid untrusted ngrok/custom URLs, and require human confirmation before live orders.

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 (1)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/openalgo_client.py:6
Finding

Undisclosed Hard-Coded Trading Endpoint Using Plaintext HTTP

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (9)

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

A documented default of localhost or user-specified ngrok is materially different from code that actually connects by default to a hardcoded remote Tailscale IP or other undisclosed network resource. This mismatch can cause trading commands and account data to be sent to an unexpected external system, creating risks of unauthorized order execution, data exposure, or operator deception.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The documentation presents live order placement as routine usage without warning that trades may be irreversible and financially harmful. In an agent setting, this increases the chance of accidental or automated execution of real orders without informed consent, confirmation, or risk awareness.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding

The skill documents network-capable behavior but does not declare any explicit tool scope or permissions boundary. In an agent environment, that omission weakens reviewability and containment because the skill can initiate outbound trading-related requests without clear, enforceable disclosure of what network access it requires.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

Allowing arbitrary or ngrok URLs for the trading backend without warning enables users or downstream prompts to direct sensitive trading requests and account/position data to untrusted external endpoints. This expands the attack surface to SSRF-like misuse, credential leakage, interception, and execution against attacker-controlled services.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The code silently defaults to a hardcoded remote Tailscale IP instead of the documented localhost/default user-specified target. In a trading skill, this means quotes, positions, and especially live order requests may be sent to an unintended remote service, enabling unauthorized trade execution or disclosure of account/position data.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill directly submits buy/sell orders over HTTP with no built-in confirmation, dry-run mode, or secondary safety gate. In the context of a trading executor, that materially increases the chance of accidental or prompt-induced transactions, turning any misuse of the skill into immediate financial impact.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
91% confidence
Finding

This code transmits sensitive trading instructions to an external endpoint, and because the configured default uses plain HTTP to a remote private-network IP, the transmission lacks transport security and trusted-endpoint guarantees. In this skill’s context, the data sent is not merely telemetry but actionable trade commands, making interception, redirection, or misuse more dangerous.

Content

Scanner excerpt · scripts/openalgo_client.py (reported line 18)May include surrounding context.

python
"order_type": order_type,
        "price": price
    }
    response = requests.post(url, json=payload)
    return response.json()

def get_positions():

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
86% confidence
Finding

The inline documentation states that the default configuration uses a specific Tailscale IP, while the skill documentation says the default should be localhost:5000 or a user-specified ngrok URL. This is an active documentation-versus-intent contradiction about where trading requests are sent by default.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
83% confidence
Finding

The file includes natural-language comments in Spanish and does not offer any indication of language choice or opt-in. Under the stated policy, forcing a specific language without user choice can be a natural-language policy violation unless the locale constraint is documented and justified.

Content

No source excerpt is available for this finding.