$ loading_
Load models and run interpretability analyses, probes, steering, and ablations.
Copy the install command and let the AI configure it · recommended for beginners
No copy-paste install info for "chuk-mcp-lazarus" yet — see the docs or source repo.
Use this MCP tool to load the specified model and extract activations from a given layer for the provided input, returning structured results for further analysis.
Returns the selected layer activations, shape metadata, and analysis-ready structured data.
Using a labeled dataset, train a linear probe with this MCP tool to detect whether the model encodes the target feature, and summarize the training results.
Outputs probe training metrics, evaluation results, and whether the target feature is captured in internal representations.
Use this MCP tool to run steering and ablation experiments on the specified layer, compare outputs before and after intervention, and summarize the differences.
Returns pre/post intervention output comparisons, magnitude of changes, and interpretability conclusions.
Enhance MCP tools with proxying, sessions, auth, storage, and progress updates.
Run mechanistic interpretability experiments and probe model features on your own compute.
Connect to Jupyter via MCP to run code and explore data interactively.
Build, debug, and manage software tasks with natural language across LLMs.
Chat with AI to retrieve documents and trigger MCP-powered tools.
Connect AI to Chrome tabs for inspection, debugging, and runtime diagnostics.