The following document will self-destruct... wait, no, it's just a GitHub README. Carry on.
Good evening, Agent. Your digital footprints betray you. Those LLM prompts you've been typing into Ollama? They're all being recorded. Every embarrassing question. Every half-baked prompt. Every time you asked the AI model to "write a love letter to my rubber duck in the style of Ernest Hemingway." Yes, that one too.
But fear not. We've developed a specialized tool for just this scenario: Operation Amnesia (codenamed "Ollama History Cleaner").
This tool will help you maintain OPSEC (Operational Security) by eliminating all traces of your questionable prompt engineering experiments.
Ollama History Cleaner is a covert Python application that infiltrates your system's Ollama data storage locations and scrubs them clean of any incriminating evidence. Like a digital janitor with top-secret clearance.
- Cross-Platform Infiltration: Deploys seamlessly on Windows, macOS, and Linux with automatic detection of target locations
- Deep Cover Operations: Identifies and neutralizes history files in their standard hideouts
- Surgical Precision: Select exactly which types of records to expunge (logs, chat history, command inputs)
- Extraction Protocol: Option to create backups before sanitizing (in case you need to restore your cover identity)
- Custom Infiltration Points: Add non-standard locations for specialized operations
- Python 3.6 or higher
- Standard issue Python libraries (no additional equipment required)
-
Clone this repository to your local machine (we recommend using a secure connection):
git clone https://github.com/Jmi2020/ClearQuestionHistory.git cd operation-amnesia -
Execute the primary operation:
python ollama_cleaner.py
-
The covert GUI interface will be established, allowing you to:
- Reconnaissance: Scan for history files to identify targets
- Target Selection: Choose which types of history to eliminate
- Execution: Clean the selected history with the push of a button
Upon deployment, the application will:
- Identify your operating system and standard Ollama data locations
- Perform reconnaissance to locate any history-related files
- Display discovered intelligence for your review
- Upon your authorization, sanitize the target files while maintaining their structural integrity
- Report mission success with detailed operation logs
The tool operates using several sophisticated techniques:
- Adaptive Targeting: Detects OS-specific paths automatically, with special handling for Windows, macOS, and Linux file system differences
- Non-Destructive Operations: Empties files rather than deleting them to maintain system stability
- JSON Preservation: Carefully cleans JSON history files by removing content while keeping their structure intact
- Asynchronous Processing: Uses threading to maintain GUI responsiveness during intensive operations
- Comprehensive Logging: Detailed mission reports at every step
- The application does not currently identify Ollama history stored in non-standard locations unless you manually specify them
- Some highly-classified Ollama configurations may use custom storage methods not covered by this tool
- This tool cannot erase memories from the minds of those who saw you typing "explain quantum physics using only emoji" into your LLM
If you encounter issues in the field or need to request additional features, establish contact through the standard communication channels (Issues and Pull Requests).
This tool is provided for legitimate purposes such as privacy maintenance, disk space management, and recovering from embarrassing prompt experiments. The developers disavow any knowledge of its use to hide evidence that you've been asking an AI to write your work emails in the style of a 1920s gangster.
This README will not self-destruct. You'll have to delete it yourself, like an adult.
END OF BRIEFING // GOOD LUCK, AGENT