Back to skill

Security audit

finance-coach-expert

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this skill suspicious because scanners found concerns about financial conclusions, saving sensitive information, and incomplete inspection; intent remains unclear.

Consider avoiding installation until the developer explains how sensitive answers are stored and addresses the confirmed incomplete-data net-worth concern. Scanner coverage was partial: 33 of 34 components were scanned, and the referenced image could not be fully inspected, so the results do not establish that the artifact is safe.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
  • 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
Findings (12)

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 23)May include surrounding context.

md
- `finance-coach-expert/avatars/expert.png`

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: file_read, file_write.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

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 · SKILL.md (reported line 26)May include surrounding context.

md
- `finance-coach-expert/avatars/expert.png`
- `finance-coach-expert/skills/asset-health-check/CHANGELOG.md`
- `finance-coach-expert/skills/asset-health-check/README_____.md`
- `finance-coach-expert/skills/asset-health-check/SKILL.md`
- `finance-coach-expert/skills/asset-health-check/references/data_format.md`
- `finance-coach-expert/skills/asset-health-check/references/dual_audience.md`
- `finance-coach-expert/skills/asset-health-check/references/health_card_template.md`

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 · SKILL.md (reported line 34)May include surrounding context.

md
- `finance-coach-expert/skills/asset-health-check/scripts/health_check.py`
- `finance-coach-expert/skills/finance-101/CHANGELOG.md`
- `finance-coach-expert/skills/finance-101/README_____.md`
- `finance-coach-expert/skills/finance-101/SKILL.md`
- `finance-coach-expert/skills/finance-101/references/ch01_mindset.md`
- `finance-coach-expert/skills/finance-101/references/ch02_inventory.md`
- `finance-coach-expert/skills/finance-101/references/ch03_products.md`

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 · SKILL.md (reported line 47)May include surrounding context.

md
- `finance-coach-expert/skills/finance-101/references/onboarding.md`
- `finance-coach-expert/skills/scam-radar/CHANGELOG.md`
- `finance-coach-expert/skills/scam-radar/README_____.md`
- `finance-coach-expert/skills/scam-radar/SKILL.md`
- `finance-coach-expert/skills/scam-radar/references/case_material.md`
- `finance-coach-expert/skills/scam-radar/references/check_entrances.md`
- `finance-coach-expert/skills/scam-radar/references/dialog_flows.md`

Net-worth conclusions are produced from incomplete data despite the documented framework-only fallback.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

The calculation produces net worth whenever either valid assets or valid liabilities exist, effectively treating the missing side as zero; invalid entries can also be excluded while a definitive total remains. In a household-finance tool, this can materially misrepresent financial health, and the general coverage disclaimer does not implement the promised framework-only fallback.

Content

No source excerpt is available for this finding.

The supposedly neutral comparison labels concentration as high and recommends reviewing diversification.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

L010 explicitly prohibits high/low judgments, and L440 describes this rendering section as factual comparisons with zero judgment. Nevertheless, L461 emits “单类占比高建议自行复核分散” whenever the concentration threshold is crossed, adding a high-concentration assessment and directional guidance rather than only reporting the percentage.

Content

No source excerpt is available for this finding.

Tainted flow: 'card' from pathlib.Path.read_text (line 924, file read) → pathlib.Path.write_text (file write)

Medium
Category
Data Flow
Confidence
65% confidence
Finding

Data from a source is assigned to a variable that is later passed to a sink, creating a variable-mediated taint flow.

Content

Scanner excerpt · finance-coach-expert/skills/asset-health-check/scripts/health_check.py (reported line 937)May include surrounding context.

python
if args.output:
        out_path = Path(args.output)
        out_path.write_text(card, encoding="utf-8")
        print(f"[OK] 体检卡已写入: {out_path}")
        if hits:
            print(f"[WARN] 合规自扫命中 {len(hits)} 处,见 stderr", file=sys.stderr)

The workflow saves user-provided financial-marketing details to case.json without a user-facing persistence warning.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The conversational workflow directs the assistant to create case.json without explaining to the user that their answers may persist on disk. Financial-marketing narratives can contain sensitive transaction or relationship details, creating avoidable exposure through shared workspaces, backups, or repository commits even without external transmission.

Content

No source excerpt is available for this finding.

The reporting workflow instructs uploading chat records and transfer receipts without a privacy warning.

Medium
Category
Output Handling
Confidence
83% confidence
Finding

Step 3 actively directs the user to upload chat records and transfer receipts to an external reporting service, rather than merely listing an available tool. These materials can contain account identifiers and other people's personal information, but the workflow gives no warning about that disclosure or guidance on limiting unnecessary sensitive information.

Content

No source excerpt is available for this finding.

The interview silently persists conversational answers and quoted statements in case.json.

Medium
Category
Output Handling
Confidence
93% confidence
Finding

The execution script tells the assistant to turn conversational answers into a persistent case.json without a user-facing storage notice or retention controls. Answers and quoted statements may reveal sensitive financial circumstances, and retaining them in an ordinary workspace unnecessarily increases access and backup exposure.

Content

No source excerpt is available for this finding.

Risk classification can understate cases with two matched features, contrary to the manifest's escalation rule.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

SKILL.md L063 promises that two or more matched features produce “高度存疑” (highly suspicious). However, L126–L128 require a weighted score of at least four or at least three matches, so two non-regulatory features such as guaranteed_profit and high_return produce only “存疑” (suspicious), understating the declared verification priority.

Content

No source excerpt is available for this finding.