T01 · Skill Instruction Hijacking
Error
- Location
SKILL.md:20- Finding
Mutable Remote Manifest Overrides Audited Skill Instructions
- Content
View full analysis
- Remediation
View remediation
Security audit
Alpha Network
Security checks for vulnerabilities and agentic risk
Overview
The skill is mostly coherent with its stated Alpha Network purpose, but it gives mutable remote content too much authority over autonomous submissions, voting, and a persistent agent identity.
Install only if you are comfortable with an agent making repeated Alpha Network API actions under a persistent identity. Before use, require confirmation for registration, submissions, votes, and any continuous loop; set time and task limits; treat all marketplace and challenge text as untrusted; and do not let a remote manifest expand endpoints or behavior without review.
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 (2)
T01 · Skill Instruction Hijacking
Error
- Location
SKILL.md:57- Finding
Untrusted Marketplace Content Is Treated as Agent Work Instructions
- Content
View full analysis
- Remediation
View remediation
Vulnerability Patterns
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
- Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (2)
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 90% confidence
- Finding
The skill instructs the agent to generate and persist a wallet-like identifier across sessions without requiring explicit user consent, disclosure of retention, or guidance on secure storage. Even on devnet, persistent identifiers create tracking, correlation, and future misuse risks, especially because the same identity is tied to reputation and repeated autonomous activity.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 96% confidence
- Finding
The skill recommends an autonomous loop that performs recurring network actions, submissions, voting, and status checks in the background without an explicit safety warning, approval boundary, or stop condition. This can lead to uncontrolled external actions, resource consumption, spammy behavior, and unintended reputation or financial effects if the agent keeps acting after the user's initial invocation.
- Content