Back to skill

Security audit

okx-sentiment-tracker

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly for OKX news and sentiment, but it handles live OKX credentials unsafely and includes a workflow that can read private account balances and positions.

Install only if you are comfortable with a global OKX CLI and live OKX credentials. Do not paste real API secrets into command-line examples; use a safer credential setup path if available, and avoid workflows that read balances or positions unless you explicitly intend to share that account data.

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 (3)

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:54
Finding

Live OKX credentials are passed through command-line arguments

Content
View full analysis
SK= PP= name=live` ``` ### Technical Analysis The profile-creation command places the API key, secret key, and account passphrase directly in command-line arguments. Command arguments are not an appropriate channel for secrets because they may be exposed through: - Shell history files. - Process inspection utilities while the command is running. - Terminal session recording. - Operating-system audit and process telemetry. - CI/CD, orchestration, or support logs that record executed commands. - Agent tool-call logs and conversational transcripts. The credentials are explicitly associated with a live profile. Although this Skill documents read-only news operations, the same credentials may be accepted by other OKX interfaces. Their effective privileges depend on how the key was provisioned and cannot be constrained by this documentation alone. ### Attack Path 1. A user follows the documented recovery instructions after receiving a demo-mode error. 2. The user substitutes real values into the `AK`, `SK`, and `PP` command arguments. 3. The shell, terminal, operating system, Agent runtime, or automation environment records the complete command. 4. A local user, malware process, log operator, or other party with access to that telemetry retrieves the credentials. 5. The exposed credentials are used against the OKX account within the permissions granted to the API key. ### Impact ...[truncated 377 chars]
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
references/workflows.md:342
Finding

Workflow accesses account balances and positions outside the Skill's declared scope

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Warning
Location
SKILL.md:41
Finding

Globally installed third-party CLI is trusted without artifact integrity verification

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (7)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger description is extremely broad, covering generic requests like 'latest news', 'market update', and 'daily briefing'. This can cause the skill to be invoked for ambiguous requests better handled by other tools, increasing the chance of overreach, unintended credentialed API use, or the web-search fallback being activated in contexts the user did not intend.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill explicitly instructs the agent to expand from the declared OKX API capability into unrestricted web search when API results are sparse. That increases the trust boundary, may expose the agent to prompt-injection or unreliable third-party content, and can cause the skill to act outside its advertised scope without explicit user consent.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The documentation states that all news and sentiment commands are read-only, but later instructs the agent to perform account-specific position analysis and give trading-style suggestions such as TP/SL or leverage changes. This inconsistency can mislead a supervising agent or reviewer about the true behavioral scope of the skill, increasing the chance of unintended sensitive-data access and unsafe financial guidance.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

This workflow expands a read-oriented news/sentiment skill into account-specific position analysis by directing use of portfolio data and generating actionable position-management suggestions. That scope creep can cause an agent to access sensitive financial account context and provide quasi-trading advice that the skill metadata explicitly says should be handled by other skills and only with clear user intent.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

The instruction mandates language choice by inference: '--lang zh-CN for Chinese, --lang en-US for English (default).' This sets locale behavior automatically rather than offering a user choice or documenting an opt-in, which can conflict with language/locale policy requirements.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The markdown explicitly frames the workflow around a Chinese-language user query (BTC 受哪些宏观数据冲击) alongside English, which signals a language-specific invocation example without clarifying that users may use any preferred language. Under the policy, forcing or privileging a specific language without opt-in can be a locale/language policy concern.

Content

No source excerpt is available for this finding.