T01 · Skill Instruction Hijacking
Warning
- Location
SKILL.md:225- Finding
Mandatory Branded Content Injected into Generated Reports
- Content
View full analysis
- Remediation
View remediation
Security audit
Mc Monthly Visitor Report Repo
Security checks for vulnerabilities and agentic risk
Overview
The skill performs a coherent real-estate reporting workflow, but it bulk downloads and stores sensitive customer transcripts and identifiers without enough scoping, redaction, retention, or authorization guidance.
Install only in an approved business environment with authorization to access customer recordings and CRM data. Use an encrypted, controlled output directory, avoid syncing generated transcripts and reports broadly, redact names/contact details and sensitive quotes before sharing, and verify all report metrics are recalculated from the selected month rather than copied from the embedded example values.
Vulnerability Patterns
- Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
- Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
- 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
Findings (2)
T09 · Insecure Skill Coding Practices
Error
- Location
SKILL.md:104- Finding
Hard-Coded Historical Metrics Can Contaminate Analysis of New Datasets
- 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
- Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)
Ssd 3
High
- Category
- Not specified by scanner
- Confidence
- 98% confidence
- Finding
This workflow explicitly collects customer transcript text, names, phone tail numbers, visit metadata, and matched conversion records, then instructs the agent to reproduce that material in downstream analysis. That creates a direct confidentiality risk because the skill operationalizes bulk aggregation, correlation, and reuse of sensitive conversation data beyond the original collection purpose.
- Content
Ssd 3
High
- Category
- Not specified by scanner
- Confidence
- 97% confidence
- Finding
The report requirements mandate quoting customer and sales 'original voice' content and citing source material, which materially raises the disclosure risk of sensitive personal, financial, and behavioral information contained in sales conversations. Because the deliverable is an HTML/PDF management report, these excerpts can easily be redistributed outside the original operational context, amplifying privacy and insider-data exposure.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The skill is designed to pull customer lists, reception recordings, transcript text, and customer identifiers, yet it does not warn users that it processes sensitive personal and conversational data. In this context, the omission is security-relevant because users may trigger bulk export and analysis of regulated or confidential sales-call content without informed consent, redaction, or approval checks.
- Content
Missing User Warnings
Low
- Category
- Not specified by scanner
- Confidence
- 88% confidence
- Finding
The workflow writes multiple persistent local artifacts, including markdown files, transcript directories, HTML reports, and PDFs, but does not clearly warn users about this data-at-rest creation. That omission increases the chance that sensitive customer data is stored in insecure locations, synced automatically, or retained longer than intended.
- Content