subagents-claude
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill functions as an orchestrator that ingests external tasks and delegates them to subagents through a
spawntool. This architecture is vulnerable to indirect prompt injection if the task input contains malicious instructions designed to manipulate subagent behavior. - Ingestion points: The agent processes a "task given with the invocation" as its primary input source (found in
SKILL.md). - Boundary markers: No delimiters or safety instructions are provided to separate the orchestrator's logic from the untrusted task content.
- Capability inventory: The skill utilizes a
spawntool to create new agent roles and assign work (SKILL.md). - Sanitization: There is no evidence of input validation, filtering, or escaping for the task content before it is passed to the tool.
- [NO_CODE]: This skill is entirely instructional and does not distribute any executable scripts, binaries, or external dependencies.
Audit Metadata