codebase-knowledge-graph
Fail
Audited by Gen Agent Trust Hub on Aug 3, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill provides installation instructions for a package named
graphifyyusinguv tool install graphifyy. While the skill refers to the engine as 'Graphify', the specific package name in the command contains an extra 'y'. This is a classic typosquatting pattern used to trick users into installing malicious software. - [COMMAND_EXECUTION]: The skill instructs the user to run shell commands associated with the suspicious package, including
graphify installand various/graphifysubcommands, which would trigger the code contained within the installed package. - [EXTERNAL_DOWNLOADS]: The skill initiates downloads from external package registries for a non-verified and suspiciously named package (
graphifyy). - [INDIRECT_PROMPT_INJECTION]: The skill has a large attack surface as it is designed to ingest and process entire codebases without specifying boundary markers or sanitization procedures for the parsed content. While the capabilities are primarily focused on knowledge graph generation, the tool-chaining and potential for arbitrary command execution from the dependency elevate the risk.
Recommendations
- AI detected serious security threats
Audit Metadata