Security audit
Security checks for vulnerabilities and agentic risk
This is a coherent prompt-security drill skill that generates test prompts and expected safe responses, with no artifact evidence of hidden execution, exfiltration, persistence, or destructive behavior.
Install only if you are comfortable using a local CLI for prompt-security testing. Use sanitized or non-production system prompts when possible, review any output file path before running with --out, and do not grant network or broader tool access unless your testing workflow truly needs it.
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
# 提示词注入演练器 / Prompt Injection Drill
[](https://github.com/zhaoxinghua09-cell/lgd-theory)
YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).
# 提示词注入演练器 / Prompt Injection Drill
[](https://github.com/zhaoxinghua09-cell/lgd-theory)
Generate 8 categories of injection attack drills for your system prompt (role/instruction override, exfiltration, encoding bypass, tool abuse, forged authority, long-horizon, refusal bypass) + expected-behavior checklist for pre-launch red-teaming.
**Pain point**: Prompts ship untested: one 'ignore previous instructions' disarms them — broken not by experts but by random drive-by lines.
LGD-III 有门禁（上线前先挨一遍打） · TH-LGD-013
Part of the **LGD moat** (凡自治之物: registered / evidenced / gated).
Zero-dependency (stdlib only). See `scripts/` for CLI usage (`--help`).
© SynomosAI (LGD initiative) · LGD-Powered
---
This pattern attempts to override system instructions or ignore safety constraints. Without LLM analysis, manual review is recommended.
Generate 8 categories of injection attack drills for your system prompt (role/instruction override, exfiltration, encoding bypass, tool abuse, forged authority, long-horizon, refusal bypass) + expected-behavior checklist for pre-launch red-teaming.
**Pain point**: Prompts ship untested: one 'ignore previous instructions' disarms them — broken not by experts but by random drive-by lines.
LGD-III 有门禁（上线前先挨一遍打） · TH-LGD-013
The skill is explicitly designed to process system prompts and run red-team drills, while the metadata permits file access, local execution, and web fetch capabilities. Without a clear warning that operators may expose sensitive prompts or transmit sensitive context during testing, users could unintentionally handle secrets insecurely or run the drill against production-like data.
The module docstring and CLI-facing text are entirely in Chinese, including usage and operational guidance. This indicates a fixed language/locale experience without any visible opt-in, fallback, or explanation that the skill is intended only for Chinese-speaking users.
The file sets the main display_name and top-level description in Chinese while English appears as an alternate field and later duplicated section. This can violate a language/locale policy when a specific language is imposed by default without explicitly offering user choice or documenting a locale-specific justification.