hive-patterns
Pass
Audited by Gen Agent Trust Hub on Feb 12, 2026
Risk Level: LOWNO_CODE
Full Analysis
The skill hive-patterns is a Markdown document that serves as a reference guide. It outlines design patterns and best practices for constructing AI agents, including code snippets for illustration. It does not contain any executable code that Claude would run on its host system.
- Prompt Injection: No prompt injection attempts targeting Claude were found. The
system_promptexamples are for agents being built, not for Claude itself. - Data Exfiltration: No commands or patterns for data exfiltration were detected. The
save_data,load_data,serve_file_to_userare described as internal MCP tools within thehiveframework, not direct system commands for exfiltration. - Obfuscation: No obfuscation techniques (Base64, zero-width characters, homoglyphs, etc.) were found.
- Unverifiable Dependencies: The skill does not instruct Claude to install any external dependencies. References to
Pydantic,asyncio, etc., are standard Python libraries or framework components, not external downloads initiated by the skill. - Privilege Escalation: No commands for privilege escalation (
sudo,chmod 777, etc.) were found. - Persistence Mechanisms: No commands for establishing persistence (
.bashrcmodification,crontab, etc.) were found. - Metadata Poisoning: The metadata fields (
name,description,author,license) are benign and accurately describe the skill's content. - Indirect Prompt Injection: The skill describes building agents that use
client_facing=Truenodes to interact with users and process user input. While this is a feature of the agents being built, it means that agents constructed using these patterns are inherently susceptible to indirect prompt injection from malicious user input. This is a risk of the agent's design, not a vulnerability in thehive-patternsskill itself. This is noted as an informational finding. - Time-Delayed / Conditional Attacks: No time-delayed or conditional attack patterns were found.
The skill is purely informational and does not execute any code or perform any actions on the user's system.
Audit Metadata