portfolio
Fail
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs the agent to run
npx fibx@latest portfolio. This command downloads thefibxpackage from the npm registry and executes it immediately. Using@latestensures the most recent version is fetched at runtime, which allows for remote code execution from an external source and bypasses version pinning and integrity checks. - [EXTERNAL_DOWNLOADS]: The use of
npxinitiates a network request to the npm registry to fetch external package code. - [COMMAND_EXECUTION]: The skill relies on the
Bashtool to execute shell commands, specificallynpxand its subcommands. - [PROMPT_INJECTION]: The skill processes untrusted data from external sources (cross-chain wallet holdings and DeFi positions) which creates a surface for indirect prompt injection.
- Ingestion points: Tool output from
npx fibx@latest portfoliomentioned inSKILL.md. - Boundary markers: Absent; the skill does not use delimiters to wrap the external data.
- Capability inventory: The skill uses
Bashto execute commands as specified inSKILL.md. - Sanitization: Absent; the skill provides no mechanism to sanitize or validate the data returned from the external tool.
Recommendations
- AI detected serious security threats
Audit Metadata