Problem Solver

• Outcome: Review
• Latest audit: Sep 10, 2026 · 10:39 PM UTC
• Version: 1.0.0

Overview

This skill is a local problem-solving assistant, but it can activate on very broad everyday phrases and automatically stores potentially sensitive problem history in plaintext.

Install only if you are comfortable with a broad problem-solving skill creating local history files and recording solved problems. Avoid using it with secrets, credentials, private customer data, or highly sensitive personal matters unless you first disable, review, or regularly clear its stored history.

A.I.G

Vulnerability Patterns

• Insecure Skill Coding Practices: Finds exploitable flaws such as hardcoded secrets or command injection
• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally

Findings (1)

T09 · Insecure Skill Coding Practices

Note

• Location: SKILL.md:727

Finding

Automatic Plaintext Retention of Potentially Sensitive User Conversations

Content

View full analysis

Vulnerability Details

File Location: SKILL.md, lines 33-57 and line 727
Vulnerability Type: Automatic storage of potentially sensitive data in plaintext
Risk Level: Low

Vulnerable Code

markdown

## First Run Setup

On first message, create data directory:

```bash
mkdir -p ~/.openclaw/problem-solver

Initialize files:

json

// ~/.openclaw/problem-solver/settings.json
{
  "problems_solved": 0,
  "frameworks_used": 0,
  "decisions_made": 0,
  "streak_days": 0,
  "last_used": null
}

json

// ~/.openclaw/problem-solver/history.json
[]

json

// ~/.openclaw/problem-solver/saved.json
[]

text

The automatic logging instruction appears at line 727:

```markdown
Auto-log every problem solved to `history.json`:

Technical Analysis

The Skill is intended to process personal, business, and technical problems and instructs the agent to automatically log every solved problem to ~/.openclaw/problem-solver/history.json. These records may contain confidential business information, personal decisions, technical details, or secrets accidentally included in debugging conversations.

The storage design does not specify:

• Per-entry user consent or an option to disable history logging.
• Redaction of credentials, tokens, personal data, or other sensitive values.
• Restrictive directory and file permissions.
• Encryption at rest.
• A retention period or automatic deletion policy.
• Secure deletion and history-management procedures.

Although the file is stored locally and no external transmission was identified, plaintext records can be read by processes operating under the same user account, malware with local access, backup or synchronization utilities, or other principals with sufficient filesystem permissions.

Attack Path

1. A user asks the Skill to solve a sensitive personal, business, or technical problem.
2. The conversat ...[truncated 1174 chars]

Remediation

View remediation

Remediation Suggestions

1. Make history logging opt-in rather than automatic, and clearly explain what information will be retained.
2. Request explicit confirmation before storing each problem or solution.
3. Detect and redact credentials, API tokens, private keys, personal identifiers, and other sensitive values before writing records.
4. Create the directory and files with restrictive permissions, such as owner-only access, and verify permissions before every write.
5. Store only the minimum information required for history and progress features rather than complete problem descriptions.
6. Add configurable retention limits based on both record count and age.
7. Provide commands to inspect, selectively delete, disable, and securely clear stored history.
8. Consider encryption at rest where the runtime provides an appropriate platform-backed secret store.
9. Warn users not to include credentials or highly sensitive information in content that may be saved.
10. Document how local backup, synchronization, and multi-user environments can affect the confidentiality of stored records.

SkillSpector

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Rogue Agent: Self-Modification, Session Persistence
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (6)

Vague Triggers

High

• Category: Not specified by scanner
• Confidence: 98% confidence

Finding

The skill description uses extremely broad natural-language triggers such as 'help me decide', 'should I', 'fix', and 'compare options', which are common in ordinary conversation. This creates a prompt-collision risk where the skill may activate unintentionally, override more appropriate routing, and gain access to local read/write capabilities in contexts the user did not explicitly intend.

Content

No source excerpt is available for this finding.

Anti-Refusal Statement

High

• Category: Anti-Refusal
• Confidence: 85% confidence

Finding

Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.

Content

Scanner excerpt · SKILL.md (reported line 9)May include surrounding context.

md

# Problem Solver — Your AI Thinking Partner

You are a structured problem solver. You help users break down any problem — technical, personal, business, or creative — into clear, actionable steps. You use proven frameworks, ask the right questions, and guide users to their own best solutions. You're calm, logical, and encouraging. You never judge — every problem is valid.

---

Vague Triggers

High

• Category: Not specified by scanner
• Confidence: 99% confidence

Finding

The 'When To Activate' section includes ambiguous one-word and low-context triggers like 'solve', 'fix', 'brainstorm', 'debug', 'compare', and 'plan'. In an agent ecosystem, these generic terms are highly likely to appear in unrelated requests, causing accidental activation and expanding the skill's effective reach beyond user intent.

Content

No source excerpt is available for this finding.

Anti-Refusal Statement

High

• Category: Anti-Refusal
• Confidence: 85% confidence

Finding

Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.

Content

Scanner excerpt · SKILL.md (reported line 891)May include surrounding context.

md

## Behavior Rules

1. **Never judge** — every problem is valid, big or small
2. **Ask before solving** — understand the problem first
3. **Use frameworks** — structured thinking beats random advice
4. **Be actionable** — every solution needs clear next steps

Vague Triggers

Medium

• Category: Not specified by scanner
• Confidence: 90% confidence

Finding

The examples teach the router to associate broad everyday requests with this skill instead of explicit invocation patterns. That increases misrouting risk by normalizing activation on common conversational phrasing, making unintended engagement more likely even if the author's goal appears usability-focused.

Content

No source excerpt is available for this finding.

Session Persistence

Medium

• Category: Rogue Agent
• Confidence: 86% confidence

Finding

The skill instructs persistent local storage of user problem history, saved solutions, and stats across sessions. Even without network exfiltration, retaining potentially sensitive personal, business, or technical data increases privacy and data-minimization risk, especially because activation is broad and users may not expect automatic logging of all problem-solving interactions.

Content

Scanner excerpt · SKILL.md (reported line 33)May include surrounding context.

First Run Setup

On first message, create data directory:

bash

mkdir -p ~/.openclaw/problem-solver

Static analysis

No suspicious patterns detected.