fleet
Pass
Audited by Gen Agent Trust Hub on Jul 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is a set of instructions for using the platform's multi-agent orchestration tools. No malicious behavior, obfuscation, or unauthorized access patterns were found.
- [EXTERNAL_DOWNLOADS]: The documentation suggests using
pnpm installwithin child sessions. This is a standard development practice using a well-known package manager. - [COMMAND_EXECUTION]: The skill provides examples for the
agent-deckCLI to control agent sessions. These commands are part of the core functionality of the agent environment. - [PROMPT_INJECTION]: The workflow involves reading output from child sessions, creating an indirect prompt injection surface. 1. Ingestion point:
agent-deck session output(SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory:agent-decksession management commands (SKILL.md). 4. Sanitization: Absent. This is a fundamental aspect of the described multi-agent coordination workflow rather than a malicious injection.
Audit Metadata