V3 Deep Integration
Pass
Audited by Gen Agent Trust Hub on Mar 1, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references agentic-flow@alpha as a primary dependency for the new integration architecture. This is a vendor-specific resource associated with the author's ecosystem.
- [COMMAND_EXECUTION]: The skill implements logic to modify the file system and manage the agent lifecycle. Specifically, the CodeCleanup class contains methods to remove legacy source files (e.g., SwarmCoordinator.ts and AgentManager.ts) to reduce code duplication, and the SystemMigration class programmatically creates agents and executes task graphs.
- [PROMPT_INJECTION]: The skill demonstrates a vulnerability surface for indirect prompt injection by ingesting data from existing systems to drive behavior. 1. Ingestion points: The SystemMigration class extracts swarm configurations, active agent definitions, and legacy tasks from the current environment (via extractSwarmConfig, extractActiveAgents, and extractTasks). 2. Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the migration logic to prevent the agent from obeying instructions within the migrated data. 3. Capability inventory: The skill has the authority to create new agents, execute complex task graphs, and delete local source files. 4. Sanitization: There is no evidence of validation or sanitization of the extracted metadata or configurations before they are passed to the agentic-flow execution engine.
Audit Metadata