hive-mind-advanced
Warn
Audited by Gen Agent Trust Hub on Mar 26, 2026
Risk Level: MEDIUMREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructions frequently utilize
npx claude-flow, which fetches and executes code from the npm registry at runtime. This creates a dependency on an external, third-party package managed outside of the trusted vendor ecosystem. - [EXTERNAL_DOWNLOADS]: The documentation includes multiple references to external resources hosted at
github.com/ruvnet/claude-flow, which are used to guide the agent's behavior and system configuration based on third-party content. - [COMMAND_EXECUTION]: The skill requires the agent to execute shell commands for initializing the hive-mind system, spawning worker swarms, and managing session states (e.g.,
npx claude-flow hive-mind init,npx claude-flow hive-mind spawn). - [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it ingests untrusted user-supplied data (objectives) to direct agent behavior and command execution.
- Ingestion points: Data enters the context via the
objectiveargument innpx claude-flow hive-mind spawncommands. - Boundary markers: No specific delimiters or instructions to ignore embedded commands within the objective are present in the skill instructions.
- Capability inventory: The system performs shell command execution (npx), file system writes (SQLite persistence), and network-based package retrieval.
- Sanitization: There is no evidence of sanitization or validation of the input objective before it is passed to the execution environment.
Audit Metadata