team-builder
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads instructions from external markdown files and interpolates them directly into a sub-agent's prompt without using boundary markers or sanitization. If a project contains a malicious markdown file in the
./agents/directory, it could override the behavior of the spawned sub-agent. - Ingestion points: Content is read from markdown files discovered in
./agents/**/*.mdand~/.claude/agents/**/*.md(referenced in SKILL.md under Configuration and Step 1). - Boundary markers: Absent; the skill uses a simple template:
"{agent file content}\n\nTask: {task description}". - Capability inventory: The skill uses the
Agenttool to spawn sub-agents withgeneral-purposecapabilities. - Sanitization: Absent; the content of the discovered files is treated as authoritative identity and rules for the sub-agent.
Audit Metadata