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.
-
Powerful local-first browser automation for AI agents with real Chromium, isolated profiles, MCP, Maqam approvals, and verifiable evidence.
npm AGPL-3.0 ★ 9 -
Aside-class browser agent — MCP · CLI · REPL. A real logged-in browser for AI agents: Keychain credential vault (secrets never in model context), enforced approval + read-only evidence mode, review evidence bundles, research/claim-drift, record→replay, and browsing-history recall.
MIT ★ 2 -
MCP server that gives coding agents evidence-based browser audio diagnostics.
npm MIT -
Evidence-Grounded Brave Browser Agent (brave-mcp) — High-performance CDP bridge and MCP automation engine for AI agents.
-
Deterministic local browser testing with Playwright, evidence, and MCP.
npm Apache-2.0 -
Web search and browser evidence for AI agents. Read pages through an MCP server with SSRF protection.
AGPL-3.0 -
Goal-driven browser exploration with Jev, compact evidence, and persistent supervisor handoff over MCP.
Apache-2.0 ★ 1 -
PRAMAN (प्रमाण) — evidence-driven verification toolkit for AI coding agents. Authored browser/command checks, strict proof gates, CLI + stdio MCP.
MIT -
Local browser-extension risk analysis engine with code evidence, exposed as a CLI and an MCP server.
pypi Apache-2.0 -
Repeatable browser checks for coding agents. JSON scenarios, local Playwright evidence, CLI + MCP.
Apache-2.0 ★ 1 -
Verified web evidence and headless browser rendering Model Context Protocol (MCP) server for coding agents
npm MIT -
webauditgithub.com/xiaojiou176-open/webauditupdated this year
Evidence-first browser automation for AI agents and operators, with recovery and MCP.
MIT ★ 1 -
Local web search and browser-assisted research for AI agents, with source provenance, saved evidence, and MCP tools.
-
Public-safe browser control surface for InnerOS local AI, project workspaces, execution lanes, evidence, and bounded physical control.
MIT -
Autonomous browser testing that turns discovered bugs into deterministic regression tests. A model may choose what to try; only browser evidence decides pass or fail. CLI + MCP server.
MIT ★ 1 -
PAW browser-use component: viewport evidence, guarded input and reusable site packs. Chrome extension, local CLI and MCP. Development preview.
MIT -
Local-first, verifiable and auditable browser execution for AI agents. WebMCP, DOM/Shadow DOM, workflows, evidence chain and benchmarks.
MIT -
Purr-Verify-MCPgithub.com/0xheycat/Purr-Verify-MCPupdated this year
Self-hosted MCP verification and VPS operator runtime for AI agents, with durable jobs, isolated builds, browser QA, deployment, and redacted evidence.
MIT -
Reverse engineer any API: discover how a website serves its data (publisher catalogs, OpenAPI, framework routes, APIs.guru, Wayback, CT, REA browser capture, jsluice, source maps, GraphQL), prove it with evidence, export OpenAPI 3.1, and generate tested T
npm MIT -
Chat with the videos you've saved: evidence-backed answers with timestamps that jump back to the source. Browser extension + MCP, self-hostable. 和你收藏的视频对话,答案带时间戳,一键回到原视频。
AGPL-3.0 ★ 108 -
Nova AI Workspace — the local browser workspace where your AI agents work. Real logged-in sessions, terminal, memory, schedule and a deep MCP tool catalog for Claude Code, Codex, Gemini, Cursor and any MCP client. Agents reproduce web/UI bugs, inspect real browser evidence, verify fixes and reuse learned site knowledge. Windows, local-first. Alpha.
NOASSERTION ★ 1 -
Local browser control and temporal visual evidence for AI agents — an MCP server that drives Chrome, continuously records it, and answers what-happened questions from retained frames, storyboards, filmstrips, and browser events
cargo ★ 3 -
Measure and improve how text reads on the web: evidence registry, browser-based CLI, workbench, extension and MCP server.
Apache-2.0 -
Monito's hosted MCP server: check a deployed URL in a real browser and get a verdict with evidence.
MIT -
Evidence-backed public risk intelligence, entity resolution, and fraud research through CLI, MCP, and browser-assisted official sources
MIT -
Transaction security for EVM chains. ShieldBot checks tokens, approvals and signatures before a wallet signs, says what it checked, and reports Unknown instead of Safe when evidence is missing. Browser extension, Telegram bot, REST API, MCP server and SDKs. Covers Ethereum, BNB Chain, opBNB, Base, Arbitrum, Polygon, Optimism and Robinhood Chain.
MIT ★ 2 -
UXMachine MCP server: measure a website in a real browser and get observations with evidence.
MIT -
Thai-first research platform connecting lawful exact-page evidence to global research through browser-native human-agent workflows.
MIT -
A local evidence workbench for researchers and students: search PDFs and notes, build cited AI context packs, and audit quotations. CLI, browser, and MCP.
MIT ★ 20 -
GitHub Action, CLI and MCP server that replays user bug reports in a real browser under the user's own conditions, proves them with browser evidence, and opens a PR with the fix and a regression test. Runs on NVIDIA Nemotron.
Apache-2.0 ★ 2 -
Devin / Cursor-style delegated coding on your own hardware. Hermes workers in Crabbox dev boxes, with MCP, browser, desktop, and PR evidence.
MIT -
JSHook MCPgithub.com/vmoranv/jshookmcpno update data
AI-powered MCP server for JavaScript reverse engineering, security auditing, browser automation and dynamic analysis. Features CDP debugging, JS deobfuscation, network interception, memory scanning, Frida instrumentation, V8 inspector, WASM analysis, and cross-domain evidence correlation.
smithery -
Self-hosted, evidence-first web research stack for any AI agent — SearXNG, extraction, browser rendering, HTTP, Python SDK, and 16 MCP tools.
MIT -
Testers.ai CARBON Test Harness Litegithub.com/jarbon/carbon-lite/blob/main/mcp/README.mdno update data
Local, MIT-licensed AI test harness with 10 MCP tools and 11 prompts for risk-based test planning, focused testing, accessibility checks, evidence capture, screenshot maps and interactive HTML quality reports. Runs over stdio in one explicitly selected project. Your coding agent supplies AI and browser execution; CARBON records observed evidence and generates reports. No hosted backend, model API key or telemetry required.
smithery -
Local desktop testing workbench for developers and coding agents. Browser/API suites, HTTP/SSE scenarios, CLI, MCP, and recorded evidence.
MIT -
Usability tests for websites, localhost and Figma flows: synthetic participants, real browser, evidence-backed reports. Works in Claude Code and Codex.
NOASSERTION -
Check quotes from the EU AI Act against the right version and date. MCP server, signed corpus releases, browser-verifiable evidence records.
Apache-2.0 ★ 1 -
Find and verify free-access LLM API routes: dated evidence, browser-direct checks, keyless configs, change alerts, and a read-only MCP server.
NOASSERTION ★ 4 -
Preview React, browser-renderable Next.js and Vue components from source. UI edit evidence for coding agents and VS Code, without starting the app.
npm MIT ★ 7 -
Local Salesforce Apex debug-log analysis: execution traces, SOQL/DML, governor limits, and evidence-linked reports. Browser and VS Code extensions, CLI, library, and MCP server.
MIT -
A shared incident rehearsal for engineers and their WebMCP browser agents. They gather evidence on one live board, challenge and revise theories in public, vote on a fix, and leave the final simulated action to a human commander.
MIT ★ 1 -
QA for any running website: an MCP server for AI hosts (Claude Code, Antigravity, Gemini CLI, Cursor) and a standalone command, with test-account logins, real-browser actions, check packs, evidence, bug reports and approved Nashmi cards
npm MIT -
AI-agent ready (MCP): ask Claude or Cursor to audit your GTM container and Google Ads conversions, with evidence for every finding. Also runs in the browser (files never uploaded) and as a CLI. Read-only.
Apache-2.0 ★ 4 -
A command driven terminal for domain and network diagnostics. 26 commands served to browsers, curl, MCP and WebMCP — an agent's tool call renders on the page, so the person reads the same evidence the agent got. Live at dug.sh
MIT ★ 1 -
SAP ABAP MCPgithub.com/Coaspe/sap-abap-mcpno update data
SAP ABAP MCP is a local-first Model Context Protocol server that enables AI coding agents to develop, test, analyze, and operate SAP ABAP systems through the official ABAP Development Tools HTTP services. It supports repository search and source editing, structured DDIC workflows, enhancement inspection, paged package navigation, source-backed object creation, semantic services, activation, ABAP Unit and ATC, transports, abapGit, RAP generation, debugging, bounded class and executable-program profiling, runtime dumps, traces and feeds, cross-system comparison, dependency analysis, guarded refactoring, and an opt-in same-origin Screen/GUI Status bridge. Curated compact, development, and assurance presets reduce tool-schema token use without removing capabilities from the default surface. The default local stdio mode can be extended with governed self-hosted Streamable HTTP, API-key roles, OIDC/JWT authentication, request-scoped SAP bearer forwarding, per-session isolation, rate limits, structured audit logs, or a side-effect-free embeddable Node.js API. It also includes browser OAuth Authorization Code with PKCE, a read-only transport assurance CLI and GitHub Action, BTP service-key profile import, and live-SAP evidence tooling. SAP profiles and credentials remain on the user's computer, with passwords and OAuth credentials protected by macOS Keychain or Windows DPAPI. SAP-dependent capabilities are validated against the user's own SAP release, configuration, and authorizations.
smithery -
Human-controlled WebMCP workspace for reviewing publishable claims against exact evidence before export.
MIT -
agent-evidence-mcpgithub.com/xiexie-qiuligao/agent-evidence-mcpupdated this year
面向 agent 长任务的本地 MCP,自动保存截图、短录屏、打码副本和可复盘证据。
MIT ★ 4 -
@nangman-infra/touch-browser-mcpgithub.com/nangman-infra/touch-browserupdated this year
Headless MCP entrypoint for touch-browser public docs and research web workflows.
npm MPL-2.0 -
AI 에이전트용 변조 증거(tamper-evident) 웹 증거 MCP 서버 — 브라우저 캡처를 SHA-256 등록하고, 인용 주장이 등록 바이트와 불일치하면 실패시키는 cite-or-fail 클레임 게이트, 오프라인 재검증 가능한 Merkle 증거 번들(.evb) 수출. Apache-2.0
npm NOASSERTION -
AI said it finished. Flyto2 shows the proof.
pypi Apache-2.0 ★ 483 -
Typed semantic filtering for AI agents: turn retrieved records, code and logs into compact evidence with source IDs. CLI · Python · TypeScript · MCP.
npm MIT ★ 1 -
Permission-aware AoA connector for authorized course-platform content: normalized knowledge, retrieval, evidence packets, local evals, and MCP tools.
★ 1 -
First open-source OpenEvidence MCP server: browser-session medical research tools for Codex, Claude Code, and MCP clients
npm Apache-2.0 ★ 49 -
Chrome-first shopping extension product family with 8 storefront apps, 1 Suite shell, a reviewer-facing release shelf, and a read-only stdio MCP surface.
MIT ★ 1 -
Unofficial OpenEvidence MCP server — query OpenEvidence from Claude Code, Codex, and any MCP client through your own logged-in browser tab (no API key). Fire-and-forget asks, a shared relay daemon for concurrent sessions, BibTeX + Crossref citations.
npm Apache-2.0 ★ 83 -
Read-oriented MCP server for controlled Vinted browsing, listing extraction, and verifiable evidence capture.
-
renderproofgithub.com/Djsand/renderproofupdated this year
Rendered web evidence for coding agents: screenshots, motion analysis, CSS animation metadata, and audit trails via MCP.
MIT ★ 1 -
GhostFirst's open layer: an MCP link-safety gate for AI agents and voice assistants (evidence before the click).
Apache-2.0 -
MCP server and Claude Code plugin. Runs an agent's code on fresh, isolated VMs, several at once, each run recorded with evidence
npm MIT -
Deterministic web-evidence engine: trace one field across raw HTML, structured data, embedded state, rendered DOM and network JSON, with exact provenance. CLI and MCP. No LLM.
Apache-2.0