ruflo

Warn

Audited by Gen Agent Trust Hub on Oct 14, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of the 'ruflo' suite (ruflo, claude-flow, @claude-flow/cli) and the 'skills' utility, which are downloaded from external sources (npm and GitHub) associated with an untrusted third-party developer.
  • [COMMAND_EXECUTION]: The skill instructs the agent to perform project initialization and system modification via shell commands like 'npx ruflo init', 'npx ruflo doctor --fix', and 'npx skills add', which grant the tool authority to alter the local development environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill functions as a multi-agent orchestration layer that ingests and processes untrusted data to coordinate agent swarms. • Ingestion points: Local project files, session memory stored in SQLite/AgentDB, and external plugin outputs. • Boundary markers: No delimiters or isolation protocols are defined in the instructions to distinguish between orchestration logic and untrusted data content. • Capability inventory: The skill possesses file-write access (creating .claude/ directories) and command execution capabilities. • Sanitization: There is no evidence of filtering or escaping mechanisms for external content being interpolated into agent prompts during orchestration cycles.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 14, 2026, 08:28 AM