Back to skill

Security audit

Id Emas Pro

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly coherent for Indonesian gold prices, alerts, portfolio tracking, and AI analysis, but it asks for broader local authority than it needs and stores user-linked financial records in plaintext.

Review before installing in a shared or multi-user environment. If you use it, restrict permissions to the fixed node scripts rather than broad filesystem access, protect the .data directory because it can contain holdings and alert thresholds, and only enable the AI tier/API key if you are comfortable sending price and history context to Moonshot/Kimi.

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 (2)

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
RHAONE_GUIDE.md:101
Finding

Recommended Configuration Grants Excessive Filesystem Access

Content
View full analysis
Remediation
View remediation
/scripts/main.js` - Explicitly supported subcommands and validated arguments 4. If direct filesystem tools are operationally necessary, scope them to `/.data` rather than the workspace or user home directory. 5. Run the Skill under a dedicated, unprivileged operating-system account or sandbox. 6. Document the exact files and directories required by each feature so administrators can apply a minimal policy. 7. Do not expose the API key or unrelated environment variables to agents or subprocesses that do not require them. ]]>

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/portfolio.js:10
Finding

Sensitive Portfolio and Alert Records Are Stored in Plaintext Without Explicit Restrictive Permissions

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
Findings (35)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
85% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · RHAONE_GUIDE.md (reported line 441)May include surrounding context.

bash
# Hapus cache lama, force scrape ulang
rm .data/cache.json
node scripts/main.js health

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The analyzer reports persistent local portfolio storage in .data/portfolio.json and per-user holdings/profit-loss tracking, while the visible description emphasizes price checks and analysis. Undisclosed financial record storage is more sensitive than a simple informational skill and raises risks of privacy leakage, unauthorized access to another user's holdings, and incorrect trust assumptions by users.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

The analyzer reports persistent local portfolio storage in .data/portfolio.json and per-user holdings/profit-loss tracking, while the visible description emphasizes price checks and analysis. Undisclosed financial record storage is more sensitive than a simple informational skill and raises risks of privacy leakage, unauthorized access to another user's holdings, and incorrect trust assumptions by users.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The analyzer reports persistent local portfolio storage in .data/portfolio.json and per-user holdings/profit-loss tracking, while the visible description emphasizes price checks and analysis. Undisclosed financial record storage is more sensitive than a simple informational skill and raises risks of privacy leakage, unauthorized access to another user's holdings, and incorrect trust assumptions by users.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The analyzer reports persistent local portfolio storage in .data/portfolio.json and per-user holdings/profit-loss tracking, while the visible description emphasizes price checks and analysis. Undisclosed financial record storage is more sensitive than a simple informational skill and raises risks of privacy leakage, unauthorized access to another user's holdings, and incorrect trust assumptions by users.

Content

No source excerpt is available for this finding.

Memory Manipulation

High
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.

Content

Scanner excerpt · scripts/history.js (reported line 57)May include surrounding context.

js
if (keys.length > MAX_DAYS * 10) {
      // hapus 20% tertua
      const toDelete = Math.floor(keys.length * 0.2);
      for (let i = 0; i < toDelete; i++) delete history[keys[i]];
    }

    await writeHistory(history);

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The README explicitly states that alert data is stored in a local JSON file and is associated with a userId, but it does not warn operators or users that this data persists on disk. In a multi-user agent environment, this can lead to unexpected retention, cross-user exposure, or accidental disclosure through backups, shared workspaces, or logs, especially since the storage is described as per-skill rather than globally isolated per user.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The document presents operator instructions in Indonesian (Panduan lengkap setup, deployment, dan operasional...) and continues in that language throughout, with no opt-in or explanation that the skill is intended only for Indonesian-speaking operators. This can violate language/locale policy when a specific language is effectively forced without user choice or explicit justification.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The guide instructs operators to allow broad exec capability for a skill whose primary purpose is checking gold prices and managing alerts. Granting unrestricted command execution materially expands the attack surface: if the skill, its prompts, or related automation are abused, an attacker could run arbitrary local commands under the OpenClaw context.

Content

No source excerpt is available for this finding.

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · RHAONE_GUIDE.md (reported line 409)May include surrounding context.

bash
# Cek lokasi skill
ls ~/.openclaw/workspace/skills/id-emas-pro/SKILL.md

# Reload
openclaw gateway restart

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
86% confidence
Finding

The skill instructs the agent to execute Node.js scripts that likely use environment access and outbound network access, but the manifest does not declare any explicit tool scope such as permissions or allowed-tools. This creates an authorization gap: the runtime capabilities are broader than what is documented or constrained, making misuse or unintended execution harder to govern and audit.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The trigger phrase "harga emas" is broad enough to match ordinary conversation, increasing the chance of accidental activation. In an exec-based skill, unintended invocation can cause unnecessary script execution, network requests, or disclosure of outputs the user did not explicitly request.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The alert setup trigger is underspecified, so casual phrases about wanting notifications could activate a state-changing workflow. Because alert creation stores per-user data and may create recurring notifications, accidental activation has persistence and privacy consequences beyond a one-time informational lookup.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
81% confidence
Finding

This JavaScript file contains natural-language content exclusively in Indonesian, including the file description and inline notes such as 'Daftar brand emas', 'Harga estimasi dari Antam + markup', and 'belum ada scraper — butuh login'. Under the policy rule for language or locale, this can be a violation when a skill implicitly forces a specific language without opt-in or documented regional justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This file contains multiple user-facing string literals in Indonesian, including recommendation labels ('BELI', 'TAHAN', 'JUAL', 'TUNGGU') and confidence/trend labels ('Tinggi', 'Sedang', 'Rendah', 'Netral'). Because the strings are hardcoded with no indication of localization, language selection, or documented region-specific scope in this file, the skill may violate language/locale policy by forcing a specific language on users.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The user-facing upgrade and availability messages are hardcoded in Indonesian ('Fitur tidak tersedia', 'memerlukan tier', 'Upgrade untuk mengaktifkan') with no indication that language selection is optional. This creates a natural-language locale policy issue because the skill appears to enforce a specific language without user opt-in or justification.

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/ai-analysis.js (reported line 8)May include surrounding context.

js
export class GoldAIAnalyst {
  constructor(config) {
    this.apiKey = config.apiKey;
    this.baseUrl = config.baseUrl ?? "https://api.moonshot.cn/v1";
    this.model = config.model ?? "moonshot-v1-8k";
  }

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

This code sends the assembled prompt content, including current price data and optional historical context, to an external API via HTTP POST. The file has no confirmation prompt, logging, print statement, or explanatory comment/docstring disclosing that user- or system-supplied analysis inputs are transmitted to a third-party service.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The system prompt explicitly instructs the model to always provide analysis in Bahasa Indonesia, which imposes a language choice without any opt-in or alternative. The policy allows locale constraints only when users are given a choice or when the restriction is clearly documented and justified as region-specific.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The notifier hard-codes the Indonesian locale (id-ID) for number and date formatting and uses Indonesian-only user-facing text. This creates a natural-language/locale policy issue because users are not offered a language or locale choice, and no justification for the locale restriction appears in the file.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The AI analysis path requires an external API key and sends price/history context to a third-party model, but the CLI does not clearly warn the user that data will leave the local environment. In this file the transmitted data appears limited to gold price/history data, so the exposure is not highly sensitive by default, but user-associated context such as user-selected brand/history usage may still create an unintended data-sharing/privacy issue.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The returned user-facing strings are in Indonesian and number formatting is forced to the "id-ID" locale. This creates a natural-language/locale policy issue because the skill does not provide any user opt-in, fallback, or documented justification for restricting output language and locale.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The handleError function returns all user-visible messages in Indonesian, including fallback and category-specific responses. This is a natural-language locale constraint with no opt-in, selection mechanism, or documentation in this file justifying that the skill is intentionally region-specific.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
86% confidence
Finding

The markdown instructs operators to run rm .data/cache.json to force a re-scrape, which deletes local cached data. While this is part of troubleshooting, the guide does not warn that the command removes cached state and may temporarily eliminate fallback data until new data is fetched.

Content

No source excerpt is available for this finding.

Vague Triggers

Low
Category
Not specified by scanner
Confidence
85% confidence
Finding

For a markdown-defined trigger, "morning brief" is a common phrase that could collide with unrelated system or human messages. The file does not constrain the source format beyond saying the system sends a message containing it, which leaves activation scope ambiguous.

Content

No source excerpt is available for this finding.