MCP servers and skills: search, filter by where they run, and compare.
1–24 of 128 tools for “prefix-cache”
Clear allRadix trie-based prompt prefix cache optimizer matching longest common prompt token prefixes to eliminate redundant prefill computation and reduce TTFT.
SkillMIT★ 7No key
Radix trie-based prompt prefix cache optimizer matching longest common prompt token prefixes to eliminate redundant prefill computation and reduce TTFT.
SkillMIT★ 7No key
High-performance OpenAI and Anthropic compatible LLM inference server for Apple Silicon. Native MLX, continuous batching, multimodal models, MCP tool calling, and Claude Code support.
PythonApache-2.0★ 1.6kNo key
The Apex MCP Gateway & Inference Router Kernel: Solving NP-Hard Bottlenecks via Alien's Trick, Radix KV-Cache, Conformal Cascading, Link-Cut Trees, SWAR Bitsets, and KKT Water-Filling.
PythonApache-2.0★ 1Few starsNo key
ctxfeed is a local MCP project-context backend that shards a whole repo into GLM-5.2's 1M-token window with cache-aware ingest ordering
PythonMIT★ 1Few starsNo key
Cost governance for Massive Intelligence (IM) agent orchestration: hard per-request, per-task, and per-day USD budgets with depletion events, pre-flight cost forecasting, prefix-cache break-even plann
Node.jsApache-2.0★ 1Few starsNo key
Stop replaying the transcript: capsule session state + rolling memory for LLM CLIs (Claude Code, Codex). Measured: −90.3% vs no-cache / −30% vs cached replay at turn 10, reproducible bench. MIT, provi
PythonMIT★ 7No key
Algorithmic Solvers for MCP Gateway Bottlenecks, Radix Prefix-Cache Routing, Model Cascades, and Multi-Agent Orchestration
PythonApache-2.0★ 1Few starsNo key
Skill для Hermes / Claude Code / Codex-агентов: автоматический выбор и настройка LLM inference engine под железо, модель и workload — MLX, llama.cpp, ExLlama, vLLM, SGLang, TensorRT-LLM.
SkillMIT★ 8No key
Python middleware that compresses and optimizes LLM context before each API call: lower token cost for apps and agents, with quality scoring and safe fallback. Wraps OpenAI, Anthropic, and Google SDKs
PythonMIT★ 8No key
Draft-target model speculative decoding verification engine with rejection sampling, acceptance rate telemetry, and dynamic speedup ratio estimation.
SkillMIT★ 8No key
Draft-target model speculative decoding verification engine with rejection sampling, acceptance rate telemetry, and dynamic speedup ratio estimation.
SkillMIT★ 8No key
Virtual paged attention KV-cache memory manager with logical block mapping, non-contiguous physical page allocation, and zero-copy fragmentation tracking.
SkillMIT★ 7No key
Virtual paged attention KV-cache memory manager with logical block mapping, non-contiguous physical page allocation, and zero-copy fragmentation tracking.
SkillMIT★ 7No key
High-precision edge and on-device LLM inference profiler calculating TTFT, TPOT, tokens/second throughput, and percentile jitter distributions (P50/P90/P99).
SkillMIT★ 7No key
High-precision edge and on-device LLM inference profiler calculating TTFT, TPOT, tokens/second throughput, and percentile jitter distributions (P50/P90/P99).
SkillMIT★ 7No key
Symmetric INT8 per-tensor and per-channel quantization engine for LLM weights and activations, with SNR and MSE reconstruction telemetry.
SkillMIT★ 7No key
Symmetric INT8 per-tensor and per-channel quantization engine for LLM weights and activations, with SNR and MSE reconstruction telemetry.
SkillMIT★ 7No key
Open-source, bring-your-own-key coding agent for the terminal. Built for cheap models: run all day on DeepSeek for a few dollars, or use OpenAI, OpenRouter, Groq, local Ollama. Explicit permissions, n
Node.jsApache-2.0★ 3Few starsNo key
AGENTIC · DESIGN — Ship agents that stay on task instead of drifting halfway through.
SkillMIT★ 7No key
A DeepSeek-native AI coding agent framework in Rust — 22 crates, CLI/TUI/HTTP API, agentic tool loop + graph retrieval + 4-layer memory + MCP.
SkillApache-2.0★ 3Few starsNo key
Assembly for LLM Agents. High-density notation for flow-state vibe coding.
SkillMIT★ 3Few starsNo key
SkillMIT★ 5No key
Universal In-Memory MCP Capability & Execution Hypervisor: Submodular Knapsack Schema Selection, Tarjan SCC Cycle Detection, Sandboxed Adapter Execution, On-Demand Repository Synthesis
PythonMIT★ 1Few starsNo key
Who stands behind it
