"Edge Impulse" matching MCP servers:

Provides an MCP-compatible profiler for measuring edge and on-device LLM inference latency, including TTFT, TPOT, tokens-per-second throughput, and P50/P90/P99 jitter distributions. It enables agents and developers to run these telemetry benchmarks from MCP clients such as Claude Desktop, Cursor, and Windsurf.

From idea to edge. Everyone has unique opinions and views of the world. Varrd makes it possible for everyone regardless of statistical, coding, or market knowledge to be able to find their unique edge. The issue with LLMs testing for edges in the market is redundant idea loops, overfit with confidence, and waste days exploring nonsense. Varrd is the infrastructure and guardrails to prevent those f

MCP server for EdgeGate that enables setting up edge-AI regression gates on Snapdragon devices directly from Claude Code, Cursor, or Claude Desktop.

An ultra-rational A2A protocol for zero-token edge pre-filtering and FEP-driven deadlock prevention. Uses Cloudflare Vectorize (384d cosine similarity) with a 24h deposit model, restricting bargaining to a 4-rally limit before forcing HTTP 402 dimension jumps.

Enables AI agents to direct, generate, and compile on-device videos with storyboards, keyframes, camera effects, and audio, then export them to Android's Gallery or Google Photos while adapting to device battery and thermal limits across Termux, ADB, or desktop.

Enables zero-dependency INT8 symmetric quantization of LLM weights and activations on a per-tensor or per-channel basis, reporting SNR and MSE reconstruction telemetry along with TTFT/TPOT latency and jitter metrics. Also exposes edge inference primitives such as PagedAttention block allocation, radix prefix caching, and speculative decoding verification through the Model Context Protocol.

Enables MCP clients to manage LLM inference memory through virtual paged-attention KV-cache block mapping, non-contiguous physical page allocation, zero-copy fragmentation tracking, radix-trie prefix caching, INT8 quantized compute, and speculative decoding verification. It also exposes prefill and decode latency telemetry so edge deployments can be benchmarked and tuned without external dependencies.

Enables edge AI agents to optimize inference memory and latency by mapping logical blocks to non-contiguous physical pages, tracking fragmentation, and exposing related MCP tools.

Enables agents to perform symmetric INT8 per-tensor and per-channel quantization of LLM weights and activations, returning SNR and MSE reconstruction metrics for edge inference optimization.

Provides MCP tools for edge-based pharmaceutical shipment disposition decisions, returning release, review, or quarantine outcomes while recording every decision in a tamper-evident, hash-chained audit trail.

Calibrated probability forecasts for any resolvable question — with evidence, prediction-market edge (Polymarket/Kalshi), and a live resolved track record.

Enables Claude to drive Google Flow in a dedicated Microsoft Edge window over CDP, firing stills and clips, tracking attempts, and downloading results.

Enables analysts to build and evaluate simultaneous VARMA models for systems of time series via exact maximum likelihood on the ATSW ladder, using univariate models as the yardstick. It guides users through loading univariate fits, identifying residual cross-correlations, estimating joint models, forecasting, impulse responses, variance decomposition, and recording decisions.

Local voice and AI platform for macOS offering speech recognition, speech synthesis, MLX-based inference, OpenAI-compatible APIs, and MCP tools for use as an assistant or integration gateway.

Enables AI-powered GitHub interactions including repository analysis, code search, PR reviews, and more through the MCP protocol.

A lightweight, air-gapped desktop cockpit connecting local LLMs (Llama 3.2) to standard MCP tools with deterministic hardware safety, acting as an edge-native MCP control plane and autonomous tool dispatcher.

Enables AI agents to delegate simple text subtasks such as summarising, translating, classifying, extracting, and reformatting to free web AIs through the user's own Edge/Chrome browser, returning only plain-text answers to save the main model's context and tokens.

Enables verification of draft-target speculative decoding with rejection sampling, acceptance-rate telemetry, and dynamic speedup estimation, alongside edge inference primitives such as INT8 quantization, paged KV allocation, and radix prefix caching.

Enables generating online webpages with a single prompt through an interactive chat interface using MCP tools.

A local multimodal recognition workbench that uses your Edge/Chrome session to access DeepSeek Web, exposing MCP tools for recognizing images, documents, audio, and video with privacy-preserving temporary processing.