firefox-devtools-mcp
Provides an MCP server that automates Mozilla Firefox via WebDriver BiDi using Selenium/WebDriver. Exposes MCP tools for navigating pages, interacting with elements by UID, capturing screenshots, handling network/console events, and optionally using privileged contexts (requires extra system access).
Score Breakdown
⚙ Agent Friendliness
🔒 Security
Runs locally and does not document transport security requirements (no explicit TLS/auth model). Privileged context and script-evaluation capabilities can materially increase attack surface; capability gating relies on environment variables/flags (not fine-grained permissions). Automation mode (Marionette/webdriver signals) can affect fingerprinting; README warns about bot detection. No explicit secrets handling guidance or rate limiting is documented.
⚡ Reliability
Best When
You have a local machine with Firefox installed, want MCP-compatible browser control for agentic browsing/testing, and can accept running an automation browser session with optional privileged capabilities.
Avoid When
You cannot run locally (no Firefox binary available), you need to avoid any form of JavaScript execution/privileged context access, or you require strict security governance over browser automation capabilities.
Use Cases
- • Browser automation and QA workflows controlled by an AI agent
- • Data collection from web pages (navigate, input, click, screenshot, inspect network/console)
- • Assistive testing of UI flows across sites using local Firefox automation
- • Agent-driven debugging by inspecting DOM snapshots, network requests, and console messages
Not For
- • Running in serverless/cloud environments (requires local Firefox and cannot run on hosting like the glama.ai example)
- • Automating untrusted targets in high-security environments without sandboxing/browser isolation
- • Use cases that require fine-grained permissioning over what the agent can do inside the browser (e.g., disabling script execution entirely)
Interface
Authentication
No user/auth scheme is documented. Instead, capability gating is done via command-line flags/env vars (not a permission model).
Pricing
Open-source (MIT) and intended for local execution; cost is primarily local compute and infrastructure for running Firefox/Selenium.
Agent Metadata
Known Gotchas
- ⚠ UID-based element actions can become stale after navigation; README recommends taking a fresh snapshot (take_snapshot) before UID tools.
- ⚠ In connect-existing mode, BiDi-dependent features (console/network events) are not available.
- ⚠ Privileged context tools require MOZ_REMOTE_ALLOW_SYSTEM_ACCESS=1; enabling them increases risk and may fail if not set correctly.
- ⚠ Windows sometimes requires invoking npx via cmd / appropriate command wrapper due to connection-closed errors.
- ⚠ Marionette must be enabled for connect-existing automation; leaving it enabled can alter fingerprint signals (bot detection risk).
Alternatives
Full Evaluation Report
Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for firefox-devtools-mcp.
AI-powered analysis · PDF + markdown · Delivered within 30 minutes
Package Brief
Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.
Delivered within 10 minutes
Score Monitoring
Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.
Continuous monitoring
Scores are editorial opinions as of 2026-03-30.