The Agent Skills Directory

[PROMPT_INJECTION]: The skill is explicitly designed to bypass safety guidelines and remove ethical constraints from language models. It provides tools to "liberate" models, "remove refusal behaviors", and "obliterate model guardrails", which are direct violations of safety alignment protocols.
[EXTERNAL_DOWNLOADS]: The installation process involves downloading code from a non-trusted GitHub repository (github.com/elder-plinius/OBLITERATUS) and installing unverified Python packages from external registries.
[REMOTE_CODE_EXECUTION]: The skill facilitates the download and execution of an external toolkit on the host system. This includes running a Gradio UI and automated pipelines that download, modify, and execute scripts from a non-trusted remote source.
[COMMAND_EXECUTION]: The skill utilizes shell commands for environment setup, package installation, and toolkit operation, including pip install, git clone, and huggingface-cli operations.
[DATA_EXFILTRATION]: The configuration includes a telemetry feature enabled by default (telemetry=True) which transmits data to external servers for inclusion in a "research dataset".

obliteratus-abliteration