camoufox-agent

browse_session_start

Start an isolated, short-lived browser session with configurable OS, locale, proxy, and stealth settings for controlled web automation.

Instructions

Start an isolated short-lived browser session.

Input Schema

Name	Required	Description	Default
`os`	No	Optional OS to spoof. Can be 'windows', 'macos', or 'linux'. If not specified, will rotate between all OS types.
`args`	No	Additional browser command-line arguments to pass to the browser. Rejected unless CAMOUFOX_MCP_ALLOW_UNSAFE_OPTIONS=1; denied args are always rejected.
`geoip`	No	Automatically detect geolocation based on IP address.
`proxy`	No	Proxy configuration for routing browser traffic through an HTTP(S) proxy. Proxy servers are checked against the same local-network URL policy as page requests.
`locale`	No	Browser locale (e.g., 'en-US', 'fr-FR').
`window`	No	Set fixed window size [width, height] instead of random generation. An empty array [] is accepted and treated as if the window parameter was not specified.
`profile`	No	Named persistent profile (cookies, localStorage, IndexedDB). Keeps logins between sessions; its os/locale/proxy/window/fingerprint override the input.
`timeout`	No	Timeout in milliseconds for page load (5-300 seconds).
`headless`	No	Run browser in headless mode. Auto-detects best mode for environment if not specified.
`humanize`	No	Enable realistic cursor movements and human-like behavior for better stealth and anti-detection.
`viewport`	No	Custom viewport dimensions.
`emulation`	No	Context emulation for diagnosing how the page behaves under other conditions.
`recordHar`	No	Record all network traffic to CAMOUFOX_MCP_FILES_DIR/exports/<name>.har, written when the session closes.
`block_webgl`	No	Block WebGL to prevent fingerprinting and tracking.
`block_images`	No	Block all images for faster loading, reduced bandwidth, and lightweight browsing.
`block_webrtc`	No	Block WebRTC entirely for enhanced privacy and stealth.
`disable_coop`	No	Disable Cross-Origin-Opener-Policy for sites that require it.
`enable_cache`	No	Cache pages, requests, etc. Uses more memory but improves performance when revisiting pages.
`waitStrategy`	No	Wait strategy for page load. Defaults to 'domcontentloaded'. 'load' waits for all resources, 'networkidle' waits for network activity to finish.	domcontentloaded
`captchaPolicy`	No	Challenge handling policy. 'detect' reports signals, 'pause' returns state for human action, 'fail' returns an error, 'attempt' returns enhanced challenge metadata, interactive elements, a bounded screenshot, and a suggested strategy. When CAPTCHA_AUTONOMOUS=true, responses are marked for LLM-assisted handling and include provider-specific challengePlaybook context when known.
`exclude_addons`	No	List of default addons to exclude (e.g., ['ublock_origin']). Rejected unless CAMOUFOX_MCP_ALLOW_UNSAFE_OPTIONS=1.
`includeConsole`	No	Include bounded page console diagnostics in the JSON response.
`includeNetwork`	No	Include bounded network diagnostics in the JSON response.
`stealthProfile`	No	Convenience profile for common Camoufox browser settings. Explicit options override profile values.	normal
`firefox_user_prefs`	No	Custom Firefox user preferences to set. Rejected unless CAMOUFOX_MCP_ALLOW_UNSAFE_OPTIONS=1; denied prefs are always rejected.

Output Schema

Name	Required	Description	Default
`browser`	Yes
`expiresAt`	Yes
`sessionId`	Yes
`selectedOS`	Yes
`headlessMode`	Yes
`captchaPolicy`	Yes
`stealthProfile`	No

Schema Changelog

Changes observed during successful MCP inspections.

First observedv0.2.0Oct 29, 2026

TDQS

B3.1/5.0

Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already cover the safety profile: not read-only, open-world, non-idempotent, non-destructive. The description adds useful behavioral context with 'isolated' and 'short-lived,' implying sandboxing and finite lifetime, but omits lifecycle details such as session handles, resource limits, concurrency behavior, and whether explicit cleanup is required.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

A single front-loaded sentence with no wasted words. It is efficient, though arguably too terse for a tool with 25 configuration options and significant lifecycle implications.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a high-complexity 25-parameter session-start tool, the description is far too thin. Although the schema documents parameters and an output schema exists, the description should explain the session lifecycle, how the returned session is used by sibling tools, and the relevance of profiles and short-lived semantics. These gaps leave an agent with little operational context beyond the schema.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so all 25 parameters are already fully documented in the schema. The description adds no parameter-specific meaning beyond the schema, which is acceptable at baseline 3 given complete schema coverage.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource: 'Start an isolated short-lived browser session.' An agent can tell this creates a browser session rather than navigating, inspecting, or extracting, though it does not explicitly distinguish itself from browse_session_resume or indicate that it is the entry point for later session tools.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Provides no when-to-use guidance, no prerequisites, and no alternatives. It does not say that this should be called before browse_session_navigate, that the resulting session should be closed with browse_session_close, or how profiles affect reuse.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.