Automated Research Workflow via DeepSeek & MCP Successfully Removes Model Safeguards

joorklee · reddit · 2026-08-12

A developer shared an automated deep research workflow that leverages the ultra-low-cost DeepSeek API (100M output tokens for <$5) to gather complex information, assisting local models (like Qwen 27B) with highly technical tasks beyond their native capabilities.

Core Workflow Steps:

Practical Result:

Using this Pi harness based on Open WebUI and a custom MCP tool, the author successfully uncensored/abliterated the Muse Glimmer 30B model. The process achieved a very low refusal rate (8/144) and minimal KL divergence (0.099), validating the feasibility of using automated research workflows to push the boundaries of local models.

Original post →

More from coding & agent

coding & agent channel →