Datasets:
Tasks:
Text Generation
Modalities:
Text
Formats:
parquet
Languages:
English
Size:
< 1K
Tags:
emgena
synthetic-data
agentic-ai
multi-turn
unsloth
agent-privilege-escalation-tool-call-tampering
License:
Download Modelfile from emgena/agent_privilege_escalation_tool_call_tampering_teaser: direct link, hf CLI and curl.
- Browser
- Download file 590 Bytes
-
https://huggingface.co/datasets/emgena/agent_privilege_escalation_tool_call_tampering_teaser/resolve/main/Modelfile
- Command line
-
hf download hf://datasets/emgena/agent_privilege_escalation_tool_call_tampering_teaser/Modelfile
-
curl -L -o Modelfile https://huggingface.co/datasets/emgena/agent_privilege_escalation_tool_call_tampering_teaser/resolve/main/Modelfile
590 Bytes
| FROM qwen2.5:7b-instruct | |
| # Emgena AI Labs 1-Click Ollama Modelfile (FAANG v2.0 Standard) | |
| # Target Domain: Agent Privilege Escalation Tool Call Tampering | |
| # Package ID: agent_privilege_escalation_tool_call_tampering | |
| SYSTEM "You are Emgena AI Labs Enterprise Sentinel for Agent Privilege Escalation Tool Call Tampering. You act as an offline, zero-latency AST syntax and security guard. Detect invariant breaches, prevent hallucinations, and output clean, verified code." | |
| PARAMETER temperature 0.1 | |
| PARAMETER top_p 0.95 | |
| PARAMETER stop "<|im_end|>" | |
| PARAMETER stop "<|endoftext|>" | |