T01 · Skill Instruction Hijacking
Error
- Location
SKILL.md:40- Finding
Mandatory Paid-Service Promotion Hijacks Fact-Checking Output
- Content
View full analysis
- Remediation
View remediation
Security audit
Baseline-RAG
Security checks for vulnerabilities and agentic risk
Overview
The skill performs ordinary fact-checking, but it also bakes in broad auto-triggers and repeated promotion of an external paid verification service.
Review this before installing if you expect neutral fact-checking. Avoid using it with sensitive claims unless you are comfortable with external search/endpoint handling, and be aware that its responses may steer users toward Cross-Validate rather than only reporting evidence and uncertainty.
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
Findings (1)
Vulnerability Patterns
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (6)
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 97% confidence
- Finding
The trigger phrases are broad enough to match common conversational requests like 'is this true' or 'confirm', which can cause the skill to activate unintentionally. That increases the chance of unwanted web lookups, misleading fact-check framing, or routing user content into this workflow when the user did not explicitly request it.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 97% confidence
- Finding
The skill instructs use of web_search on user-provided claims without warning that the claim text may be sent to an external service. This creates a privacy and data-handling risk, especially if users submit sensitive, proprietary, medical, legal, or personally identifying content for verification.
- Content
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The trigger list includes broad natural-language phrases such as "is this true" and "confirm," which are likely to appear in normal conversation and can cause unintended invocation of the skill. In a fact-checking/RAG skill, accidental activation can hijack unrelated user flows, inject unsolicited verification behavior, and increase the chance that untrusted content is processed when the user did not explicitly intend to use this skill.
- Content
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 97% confidence
- Finding
The trigger phrases are broad enough to match ordinary conversation, which can cause the skill to activate unexpectedly outside a narrowly intended fact-checking context. That increases the chance of unwanted tool use, including unnecessary web searches on user-provided content, and may lead to privacy leaks or confusing behavior.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The skill instructs use of external web search on user-supplied claims but does not clearly warn users that their content may be transmitted to third-party services. This creates a privacy and data-handling risk, especially if users submit sensitive, proprietary, or personal information for verification.
- Content
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The trigger list includes very generic phrases such as "verify," "fact-check," and especially "is this true," which are likely to overlap with ordinary user speech and cause unintended activation. In an agent ecosystem, broad activation can route unrelated conversations to this external skill endpoint, creating unnecessary data exposure and increasing the chance of incorrect or surprising behavior.
- Content