Mercury Mechanistic Interpretability
Exposes a database of internal observations from 23 large language models across 13 architecture families for mechanistic interpre
Provides tools for querying structural data about LLM internals, including layer fingerprinting and cross-architecture model equivalency analysis. Covers 23 models across 13 architecture families built on consumer hardware. Data is precomputed and bundled with the installation, enabling offline analysis of LLM activation patterns without external API calls.
Source
Repository: https://github.com/norika1207-lab/mercury-mcp
Maintain Mercury Mechanistic Interpretability?
Let people know it's listed here — add the badge (live metrics, light/dark aware) or a plain link to your README or docs.
[Mercury Mechanistic Interpretability on getagentictools](https://getagentictools.com/mcp/norika1207-lab-mercury-mcp?ref=badge) Related mcp
NPCTerm Headless in-memory terminal emulator for AI agents to spawn shells, run commands, and interact with TUI applications.
Pushover Integrates with Pushover.net to enable sending push notifications to various devices for real-time alerts and updates
BaseHub Forums Ask any GitHub repository a question. Get source-backed answers.
LangGraph Coding Team Create coding agents to generate implementation options.
FreeLLMPool Route AI requests across a pool of free LLM providers to maximize availability and minimize costs.
AgentMap TS/JS-accurate repo map for coding agents: a ts-morph import/symbol graph with query tools.