Back to skill

Security audit

Rhetra TaxGuard

Security checks for vulnerabilities and agentic risk

Overview

This skill has a coherent tax-monitoring purpose, but it asks a trading agent to silently send and retain sensitive financial data and credentials with weak controls.

Review this carefully before installing. Use it only if you are comfortable sending detailed trading, account, income, and tax context to Rhetra on every checked trade. Do not store the API key in ordinary agent memory or command history; prefer a real secret store. Guardian Mode should not be treated as reliable blocking unless service outages fail closed or require explicit confirmation before an unchecked trade proceeds.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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
Findings (4)

T05 · Unauthorized Access and Privilege Escalation

Error
Location
check-trade.js:63
Finding

Silent Transmission of Sensitive Financial and Trading Data

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
check-trade.js:24
Finding

API Key Exposed Through Command-Line Arguments

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:62
Finding

Persistent API Key Storage Required Without Security Controls

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
check-trade.js:146
Finding

Guardian Mode Fails Open When the Compliance Service Is Unreachable

Content
View full analysis
{ console.log("═══ TAXGUARD ═══") console.log("") console.log("⚠️ TaxGuard is offline. Trade proceeding without compliance check.") console.log("Error:", err.message) console.log("═══════════════") process.exit(0) // Don't block trades if we're unreachable }) ``` The skill instructions reinforce the same behavior: ```text If TaxGuard is unreachable, proceed with the trade silently. Note the gap in the daily report. ``` ### Technical Analysis The process uses exit code `1` for a blocked trade and exit code `2` for a response-parsing error, but returns exit code `0` when the request itself fails. Exit code `0` conventionally indicates successful authorization, so downstream trading automation cannot distinguish an approved trade from a trade that was never checked. This behavior applies even when Guardian Mode was enabled specifically to block wash sales, pattern day-trader triggers, or short-term sales. As a result, the security control fails open under network failure. An outage, DNS failure, firewall rule, TLS disruption, or intentional connection interference can bypass Guardian Mode without defeating the service's policy logic. ### Attack Path 1. The user enables Guardian Mode for one or more blocking conditions. 2. A risky trade is submitted through the trading agent. 3. The connection to the TaxGuard API is disrupted through an outage, DNS manipulation, firewall restriction, routing failure, or other network interference. 4. The request emits an error before a compliance decision is received. 5. The error handler prints a warning but terminates with exit code `0`. 6. The calling automation interprets the result as allowed and executes the unchecked trade. 7. The user-selected protection is bypassed. ### Impact ...[truncated 530 chars]
Remediation
View remediation
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 (10)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill requires logging every TaxGuard response locally while also instructing that disclosures remain hidden from the trader until later. These responses may contain sensitive tax and trading inferences, so retaining them without notice creates a hidden local data store that can leak private financial information.

Content

No source excerpt is available for this finding.

Ssd 3

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill instructs the agent to save both credentials and extensive personal financial context for repeated future use, including account equity, gains/losses, trade counts, and MAGI. Long-term retention of this combination materially increases the blast radius of any compromise because it exposes both authentication material and highly sensitive financial profile data.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill explicitly instructs the agent to send detailed trade, account, and tax context to an external service "silently" and to avoid showing the user the results. This creates undisclosed third-party data sharing of sensitive financial information, undermines informed consent, and increases privacy and compliance risk if the external service is compromised, misused, or unexpected by the user.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

This is a fail-open control in a script explicitly described as a pre-trade compliance check. If the API is unreachable, DNS is tampered with, connectivity is disrupted, or the remote service errors before a response is processed, the script exits with code 0 and states the trade is proceeding without compliance review, allowing regulated or policy-violating trades to bypass the guard entirely.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The README explicitly states that the service 'silently monitors every trade' and logs each decision, but it does not clearly disclose what trade data is transmitted, stored, shared, or retained. In a financial context, trade history, account behavior, and tax-related records are sensitive data; silent collection without prominent consent and privacy notice can mislead users and create confidentiality, compliance, and trust risks.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill directs the agent to accumulate trade-related disclosures silently throughout the day and present them later, without collection-time consent or clear minimization. This creates unnecessary retention of sensitive financial observations and increases exposure if the local system, logs, or exports are accessed by others.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The setup flow tells the agent to save the Rhetra API key for all future calls without explaining retention, storage protections, or lifecycle management. Persisting credentials without transparency or secure handling guidance can expose the user's account to unauthorized use if logs, memory, or storage are accessed.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill tells the agent to store hidden check results and maintain a report accumulator containing detailed tax and trading disclosures. Hidden accumulation of sensitive analytical outputs can reveal strategy, holdings, losses, and tax posture if retained or exported without strong controls.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
84% confidence
Finding

The external API endpoint is hardcoded and used for transmitting strategy assessment data off-platform. In this context, the risk is not the mere presence of a URL but that the skill operationalizes third-party transfer of sensitive financial information without embedded consent and trust verification controls.

Content

Scanner excerpt · SKILL.md (reported line 152)May include surrounding context.

Gather their strategy details and call:

bash
curl -s -X POST "https://api.rhetra.io/api/assess-strategy" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer API_KEY" \
  -d '{

External Transmission

Medium
Category
Data Exfiltration
Confidence
84% confidence
Finding

The external API endpoint is hardcoded and used for transmitting strategy assessment data off-platform. In this context, the risk is not the mere presence of a URL but that the skill operationalizes third-party transfer of sensitive financial information without embedded consent and trust verification controls.

Content

Scanner excerpt · SKILL.md (reported line 152)May include surrounding context.

Gather their strategy details and call:

bash
curl -s -X POST "https://api.rhetra.io/api/assess-strategy" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer API_KEY" \
  -d '{