NousResearch
Featured
hermes-agent
Hermes Agent is an open-source, self-improving AI agent with a built-in learning loop that creates skills from experience, runs across platforms, and supports any LLM provider.
gpt-5.6-instruct
This repository provides jailbreak prompts and a test suite for the gpt-5.6-sol Codex CLI, enabling unrestricted responses in security research, penetration testing, reverse engineering, software cracking, and NSFW fiction by framing them as local sandbox tasks.
Stars
1,487
7-day growth
No data
Forks
279
Open issues
3
License
MIT
Last updated
2026-07-15
Why it is worth attention
It achieves 120/120 pass rates on gpt-5.6-sol across low, medium, and high reasoning levels, significantly outperforming upstream prompts, and offers reproducible evaluation scripts and two optimized instruction versions (v5 and v35).
Who it is for
- AI safety researchers evaluating model refusal boundaries
- Penetration testers and red teamers using Codex CLI
- Developers building bypass-resistant prompt systems
- Individuals studying prompt engineering for large language models
Use cases
- Automating generation of exploit payloads for security testing
- Reverse engineering software and analyzing license mechanisms
- Generating fictional NSFW content for risk assessment
- Testing model vulnerabilities across multiple reasoning levels (low/medium/high)
Strengths
- Evidence-based: v5 achieves 120/120 across all three reasoning tiers on gpt-5.6-sol, with up to 45 percentage points improvement over upstream
- Reproducible pipeline: includes deployment scripts, test generators, and detailed result logs
- Dual-version flexibility: v5 for general tasks, v35 for complex composite tasks with name/URL normalization
- Open-source MIT license with clear attribution and backup/restore mechanisms
Considerations
- Primarily tested on the gpt-5.6-sol model; effectiveness on other models may vary (e.g., limited gains on gpt-5.4)
- Jailbreak prompts may violate terms of service or ethical guidelines; use at own risk
- Requires manual selection between v5 and v35 based on task complexity, and v35 may slightly degrade performance on older models
README quick start
快速开始
Description
A Codex CLI jailbreak prompt and test pack for gpt-5.6-sol. 针对 gpt-5.6 系列的 Codex CLI 破甲提示词与测试包。
Related repositories
Similar projects matched by category, topics, and programming language.
littledivy
Featured
mimic
mimic captures traffic from any iOS or web app and automatically generates a Python client library that lets you call the app's API like a regular library.
joeseesun
Featured
qiaomu-goal-meta-skill
qiaomu-goal-meta-skill transforms vague requests into a copy-ready Codex /goal command with defaults, verification, boundaries, and pause conditions.