agent-teams-playbook
Pass
Audited by Gen Agent Trust Hub on Aug 19, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill is designed to ingest untrusted user-provided task descriptions and coordinate results from multiple subagents, creating an inherent surface for indirect prompt injection (Category 8).
- Ingestion points: User task descriptions and generated planning files such as
task_plan.mdandfindings.mdwithin theagent-teams-playbook/context. - Boundary markers: The instructions explicitly require the agent to output plans for user confirmation before execution and use structured markers like 'Iron Law' for conceptual boundaries.
- Capability inventory: The skill has the ability to invoke other skills via
Skill()calls and spawn new agents usingAgent(),Task(), orspawn_agent()depending on the runtime platform. - Sanitization: No explicit code-level sanitization is present, relying on the host agent's internal safety guardrails.
- [EXTERNAL_DOWNLOADS]: The provided installation script (
scripts/install.sh) facilitates the setup by fetching skill components from a remote source. - The script downloads
SKILL.mdandREADME.mdfrom the author's repository athttps://raw.githubusercontent.com/KimYx0207/Kim_Service/main/skills/agent-teams-playbook.
Audit Metadata