MCP servers that ship as code — a repository or a package you install and run locally (usually over stdio). They have no hosted endpoint, so we cannot probe or grade them: instead each one shows its repository, install command, license, stars and when it was last updated.
Collected from GitHub repositories and npm, PyPI, Cargo, Go, Maven, NuGet and RubyGems packages tagged as MCP servers, the official MCP Registry, Docker's MCP catalog, ModelScope, Baidu MCP World, Cline and other public directories; one entry per repository. A server that also offers a hosted endpoint is listed under Remote instead. These entries are not counted in our reachability, verification or readiness statistics.
-
Debug and fault-test any WebRTC session — Chrome extension, MCP server for AI coding agents, npm library, and Playwright plugin. Live RTCPeerConnection/WebSocket inspection, reconnect testing, network-loss simulation.
MIT ★ 4 -
MCP server that lets AI agents debug kernels through QEMU's gdbstub - no gdb binary. Breakpoints, registers, DWARF lines, backtraces, and triple-fault diagnosis.
MIT ★ 1 -
Fork failed Python programs into tested alternate outcomes, with original-run controls and guarded replay.
NOASSERTION -
Framework-agnostic live WebRTC inspection and fault injection: peer connections, tracks, data channels, WebSockets, fake mic/camera, and real transport/signaling fault injection for reconnect testing.
npm MIT -
Turn C/C++ crashes into source-verified execution-flow reports. Local tracing, cross-service examples, CLI and read-only MCP.
Apache-2.0 ★ 1 -
Replay the crash. Prove the fix. 81ns in-process flight recorder, 61µs deterministic replay, and 5.9ms variant fix verification for journal-driven services and AI coding agents.
Apache-2.0 ★ 2 -
Give AI hands to touch silicon — an MCP server for SEGGER J-Link, OpenOCD and Black Magic probes. Flash, halt, step, stream RTT logs and decode ARM Cortex-M faults. Verified against real hardware.
npm MIT ★ 43 -
A flight recorder for AI agents: records every step, ranks the step most likely at fault, then proves it by replaying the run with only that step repaired, at zero model calls. Top-1 0.60 on unseen failures against 0.25 baselines. MCP server, CI regression firewall, on PyPI.
MIT