pour
OfficialServer Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| POUR_MCP_ROOT | No | The directory that MCP file paths are restricted to. Defaults to the directory the server was started in. | |
| PUPPETEER_EXECUTABLE_PATH | No | Path to a Chrome/Chromium/Edge binary for puppeteer-core to use instead of the installed browser. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
pour audits pages against WCAG 2.2 with its own engine in headless Chrome. Results depend on the page state: the viewport width (responsive pages serve different content per breakpoint), whether the page was scrolled first (lazy content), and what is open on it. Say which you used when you report a figure. Findings under "needsReview" are the engine abstaining rather than guessing: report them as needing a look, not as failures.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
| logging | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| pour_auditA | Audit a URL against WCAG 2.2 A and AA (or, with wcag3, the requirements of the WCAG 3.0 draft) with the pour engine in headless Chrome: the page is loaded, the engine runs against the live DOM, and every failing rule comes back with its severity, its success criteria, the failing elements as CSS paths with the fix, and the count of rules the engine would not judge without a human. Works on a running dev server or any URL. Results depend on the page state: the viewport width (responsive pages serve different content per breakpoint), whether the page was scrolled first (lazy content), and what is open on it. Say which you used when you report a figure. |
| pour_focus_orderA | The sequential keyboard focus order of a page, as the Tab key visits it: every reachable control in order with its tag, role, accessible name (the engine's own computation) and CSS path. Controls with a positive tabindex come first and carry the value, since it forces their position. "unreachable" lists native controls a tabindex of -1 has taken out of the keyboard's reach. The same walk as the Focus Order lens in the pour extension. Results depend on the page state: the viewport width (responsive pages serve different content per breakpoint), whether the page was scrolled first (lazy content), and what is open on it. Say which you used when you report a figure. |
| pour_screenshotC | A PNG of the page as loaded, or as seen through one of pour's simulations, the same ones the extension ships. Vision: cataract (Clouded Lens), presbyopia (Near-Vision Loss), lowAcuityMild (Slight Defocus), lowAcuity (Uncorrected Focus), lowAcuityStrong (Significant Defocus), lowAcuityHeavy (Severe Defocus), lowLight (Dim Environment), lowContrast (Reduced Contrast), retinitisRing (Ring Loss), glaucoma (Tunnel Vision), glaucomaLarge (Tunnel Vision (Large)), macularDegeneration (Central Vision Loss), macularDegenerationLarge (Central Vision Loss (Large)), diabeticRetinopathy (Patchy Vision), nystagmus (Involuntary Eye Movement), hemianopiaLeft (Left Field Loss), hemianopiaRight (Right Field Loss), amblyopia (Reduced Acuity (One Eye)), afterimages (Afterimages), ageSlider (Age Slider), scotopicRose (Rose Tint), scotopicYellow (Yellow Tint), scotopicAqua (Aqua Tint), protanopia (Red Absent (Protanopia)), deuteranopia (Green Absent (Deuteranopia)), tritanopia (Blue Absent (Tritanopia)), achromatopsia (Monochromacy), protanomaly (Red Deficiency), deuteranomaly (Green Deficiency), tritanomaly (Blue Deficiency). Sensory, motor and structure: lightSensitivity (Light Sensitivity (Metaphor)), colourHypersensitivity (Colour Hypersensitivity (Metaphor)), motionSensitivity (Motion Sensitivity (Metaphor)), hyperfocusTunnel (Hyperfocus Tunnel (Metaphor)), attentionFragmentation (Attention Fragmentation (Metaphor)), peripheralDistraction (Peripheral Distraction (Metaphor)), detailFixation (Detail Fixation (Metaphor)), processingDelay (Processing Lag (Metaphor)), sensoryInterference (Sensory Interference (Metaphor)), sensorySpike (Sudden Sensory Spike (Metaphor)), dyslexiaVisualStress (Visual Stress (Pattern Glare) (Metaphor)), dyslexiaCrowding (Crowding Effect), dyslexiaTrackingLoss (Tracking Loss (Metaphor)), dyslexiaWashout (Letter Instability (Metaphor)), dyslexiaContrastSensitivity (Contrast Sensitivity), handTremor (Hand Tremor), handTremorStrong (Hand Tremor (Strong)), restingTremor (Resting Tremor), ataxicDrift (Ataxic Drift), pointerSpasm (Sudden Jerk), pointerHidden (Hidden Pointer (Keyboard Only)), fingertipTouch (Fingertip Touch), forcedColours (Forced Colours (Windows Contrast Theme)), screenMagnifier (Screen Magnifier (400%)), textSpacing (Text Spacing), focusOrder (Focus Order), landmarkMap (Landmarks & Headings). Omit the filter for the page as it is. Useful for what the numbers cannot show: colour-vision and low-vision rendering, and the focusOrder and landmarkMap lenses drawn over the page. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 3 tools
The three tools target clearly different outputs: a WCAG rule audit, a keyboard tab-order walk, and a rendered PNG. The one soft overlap is that pour_screenshot exposes a 'focusOrder' lens while pour_focus_order returns structured tab-order data, and landmarkMap has no dedicated tool, so an agent may occasionally hesitate over which to reach for.
All three names share the pour_ prefix and snake_case, with a stable noun/action describing the artifact produced (audit, focus_order, screenshot). No mixed conventions or vague verbs.
Three tools is compact but each maps to a genuinely distinct inspection mode of an accessibility engine, and the huge simulation list lives inside one parameter rather than being split into dozens of tools. Slightly thin for the breadth of the domain, but well scoped rather than padded.
The surface covers the core workflow: rule-based auditing, keyboard reachability/order, and visual simulation rendering. Minor gaps exist around structured landmark/heading output (only available as a screenshot lens) and no way to exercise interactive states beyond page-load parameters, but nothing that blocks typical agent use.