Reconstruct
Server Details
Captured web interfaces, screenshots, flows and structured evidence for AI coding agents.
- Status
- Healthy
- Uptime
- 100.0% over the last 24 hours
- Last Tested
- Transport
- Streamable HTTP · MCP 2025-11-25
- URL
- ChatGPT
- –
- Claude
- –
TDQS
Scored across 70 tools
There are several near-duplicate families: eight-plus image getters (get_page_image, get_capture_image, get_stage_image, get_scenario_initial_image, get_replay_step_image, get_click_image, get_page_evidence_image, get_component_image) differ only by an implicit key or capture type, and asset listing is split three ways (list_assets, list_product_assets, list_archived_assets) while component discovery is spread across get_page_outline, get_page_regions, get_page_components, list_component_candidates and get_component. The descriptions do add routing hints ('prefer get_page_outline', 'prefer this over per-page list_assets'), which keeps it from being a 1, but an agent must read carefully to avoid misselection.
Nearly everything follows snake_case verb_noun (list_*, get_*, create_*, find_*, wait_*), so the pattern is predictable. Minor deviations exist where the same concept gets different verbs (list_component_candidates vs get_page_components; get_sizing/get_spacing as noun-ish getters) and singular/plural pairs like get_element_asset vs get_component_assets.
Seventy tools is far beyond what an agent can hold in view and well past the 25-tool threshold, with many redundant getters for screenshots, assets and component data. The domain (captures, pages, components, scenes, scenarios, replays, requests, skills) is broad enough that some size is justified, but this surface is clearly over-expanded.
The read/inspection side is exhaustive — versions, platforms, routes, pages, components, styles, typography, spacing, motion, assets, fonts, themes, scenes, scenarios, replays, evidence — and the request lifecycle has create/get/list/wait. Write operations are limited to create_product_request, but that appears to be the server's intended scope, leaving no obvious dead ends.
Publisher details
- Operator
- Cryptify Corporation · Publisher source
- Operator website
- https://cryptify.tech · Publisher source
- Vendor relationship
- First-party · Publisher source
- Documentation
- https://reconstruct.dev/docs/connect-ai-agents · Publisher source
- Privacy policy
- Not provided
- Trust center
- Unknown
- Restrictions
- Requires a Reconstruct account and OAuth sign-in or a personal MCP access token. Capture availability and usage quotas depend on the account’s access rights and subscription plan. · Publisher source
Available Tools
70 toolscreate_product_requestCreate product requestAInspect
Submit a product capture request for the authenticated user. requestType=add requires a public HTTPS url and an empty product; subject is optional. requestType=update requires an existing published product key; backend derives its name and source URL. Select platform/environment from get_request_options. Creates a pending request, not an immediate capture. Subscription, URL safety and duplicate checks apply. Call only when the user asks to submit a request; do not automatically submit while browsing missing captures.
Inputs: platform (required), environment (required), requestType (required), url, +4 more. Returns: Authenticated owner request including review and capture progress
find_elementsFind elementsBRead-onlyIdempotentInspect
Find visible elements by text, content, label, href, action, CSS path, or tag and return layout, effects, rectangles and computed styles.
Inputs: page (required), view (required), product (required), version (required), +5 more. Returns: Structured result wrapping a non-object value. The text content holds the same…
get_application_manifestGet application manifestARead-onlyIdempotentInspect
Return the application inventory for a captured version: route count, a page of routes with their views and desktop/mobile availability (offset/limit, default 50), stage, scenario and replay names. Page through routes with offset or use list_pages.
Inputs: product (required), version (required), limit, offset, +1 more. Returns: Capture application inventory
get_capture_assetGet capture assetARead-onlyIdempotentInspect
Materialize an archived resource or stored screenshot/MP4/WebM by relative path. Returns path, URI, hash, MIME and size without embedding bytes. Discover resources with list_archived_assets/get_fonts and recordings with get_page_artifacts. Full page JSON is excluded.
Inputs: path (required), product (required), version (required), platform (required), +1 more. Returns: Archived media materialization
get_capture_imageGet capture imageARead-onlyIdempotentInspect
Read one allowlisted WebP from a product version by relative path (for example before.webp, capture/pages/home/desktop/viewport.webp). Only .webp paths present in storage are readable.
Inputs: path (required), product (required), version (required), platform (required), +1 more. Returns: Metadata for the image in content[contentIndex]; image bytes are not duplicated…
get_click_imageGet click imageBRead-onlyIdempotentInspect
Return the before.webp or after.webp screenshot from a CLI click capture. Pass kind=before or kind=after.
Inputs: kind (required), product (required), version (required), platform (required), +1 more. Returns: Metadata for the image in content[contentIndex]; image bytes are not duplicated…
get_componentGet componentARead-onlyIdempotentInspect
Return one component candidate and its descendants. Prefers the id/parent tree; falls back to rectangle containment. Includes a crop rectangle for the page image.
Inputs: page (required), view (required), product (required), version (required), +3 more. Returns: Root element and bounded descendant evidence
get_component_animationsGet component animationsARead-onlyIdempotentInspect
Read captured motion and active CSS animation for one component ID. Paginated and optionally limited to a y range; excludes idle animation declarations.
Inputs: page (required), view (required), product (required), version (required), +6 more. Returns: Paginated component motion evidence
get_component_assetsGet component assetsARead-onlyIdempotentInspect
Read media references and geometry for one component ID. Paginated and optionally limited to a y range; resolve bytes via get_element_asset (URI/path by default).
Inputs: page (required), view (required), product (required), version (required), +6 more. Returns: Paginated component media sources
get_component_imageGet component imageARead-onlyIdempotentInspect
Return a WebP cropped to one component candidate's visible rectangle from the saved page screenshot.
Inputs: page (required), view (required), product (required), version (required), +2 more. Returns: Metadata for the image in content[contentIndex]; image bytes are not duplicated…
get_component_stylesGet component stylesARead-onlyIdempotentInspect
Read grouped layout and paint styles for one component ID, with counts and sample element IDs. Paginated and optionally limited to a y range. Set expanded=true for individual elements and rectangles; excludes text, animation, and media source bytes.
Inputs: page (required), view (required), product (required), version (required), +7 more. Returns: Paginated individual styles or deduplicated style groups
get_component_textGet component textARead-onlyIdempotentInspect
Exact follow-up: visible own text, line breaks, IDs, parents, and rectangles for one component ID. Prefer get_section_bundle text/texts first; call this when the bundle marks textTruncated or you need uncapped runs/lines. Paginated and optionally limited to a y range; no CSS or asset payloads.
Inputs: page (required), view (required), product (required), version (required), +6 more. Returns: Paginated own text of captured component descendants
get_component_treeGet component treeARead-onlyIdempotentInspect
Read a component's compact parent/child skeleton by element ID and depth, without text, CSS, animation, or asset data.
Inputs: page (required), view (required), product (required), version (required), +5 more. Returns: Paginated component ancestry and descendants
get_component_typographyGet component typographyARead-onlyIdempotentInspect
Read grouped text font styles for one component ID with counts and sample element IDs. Paginated and optionally limited to a y range; use get_fonts for family names and origin URLs.
Inputs: page (required), view (required), product (required), version (required), +6 more. Returns: Paginated deduplicated typography
get_element_assetGet element assetARead-onlyIdempotentInspect
Resolve one captured element asset. Returns path/URI, hash, MIME and byte size by default. Download or copy it outside model context. includeContent=true requests inline bytes within the configured raw-byte limit; oversized assets remain references.
Inputs: page (required), view (required), product (required), version (required), +3 more. Returns: Chosen element media reference
get_elementsGet elementsARead-onlyIdempotentInspect
Read a paginated slice of captured elements: id/parent tree, layout, effects, text/content, href/action, media source and rectangles (about 1 KB each). detail=full adds computed styles and state probes (about 10 KB each); for one element prefer get_component_styles.
Inputs: page (required), view (required), product (required), version (required), +4 more. Returns: Paginated captured element records
get_flowGet user flowBRead-onlyIdempotentInspect
Read one user flow with observed outcomes, inferred expectations and scoped evidence.
Inputs: flowId (required), product (required), version (required), environment. Returns: flow, gaps, guidance
get_fontGet fontARead-onlyIdempotentInspect
Resolve one font by family name or origin URL from get_fonts. Returns family plus origin download URLs. Original bytes are stored in the capture; read resources[].path with get_capture_asset.
Inputs: path (required), product (required), version (required), platform (required), +1 more. Returns: Original URLs and archive references for a captured font
get_fontsGet fontsARead-onlyIdempotentInspect
List font families and weights. includeUrls=true returns origin webfont URLs and archived resource paths; includeFaces=true includes face ranges. Prefer get_capture_asset(resources[].path) for stored bytes; origins are metadata and can become unavailable.
Inputs: product (required), version (required), platform (required), limit, +5 more. Returns: Captured fonts with optional faces and original URLs
get_interface_sceneGet platform-neutral interface sceneARead-onlyIdempotentInspect
Export a platform-neutral reconstruct.scene.v1 interface: profiles, frames, node hierarchy, logical geometry, neutral layout constraints, text/paint, resource references, actions, transitions, motion and explicit evidence limitations. Returns a compact artifact manifest. Consume the JSON programmatically. CSS, DOM and Android classes are adapter inputs, not canonical layout. Request each saved view separately; never infer fixed sizing from measured bounds or successful interactions from a declared action. Native captures require a published scene; do not invent them.
Inputs: page (required), view (required), product (required), version (required), +2 more. Returns: Manifest for a platform-neutral scene and archived resources
get_languagesGet languagesARead-onlyIdempotentInspect
Return languages detected for a product version on a platform (from languages.json, with path-prefix fallback). Each language lists its code, sources and urlCount; pass includeUrls=true for the full URL lists.
Inputs: product (required), version (required), platform (required), environment, +1 more. Returns: Detected locales and evidence sources
get_page_artifactsGet page artifactsARead-onlyIdempotentInspect
List stored media for one page/view: viewport and full-page screenshots, section tiles, scroll and interaction recordings. Returns relative paths and MIME types plus capture-limit metadata. Retrieve a chosen file with get_capture_asset; absent recordings are not invented.
Inputs: page (required), view (required), product (required), version (required), +1 more. Returns: Stored media and capture limitations for a page/view
get_page_bundlesGet page bundlesARead-onlyIdempotentInspect
Batch IR: multiple get_section_bundle results in one RPC. Pass sectionIds and/or regionIndexes as comma-separated lists (max 8 total); with neither, the first 3 viewport regions are returned. Prefer outline regions (bundleTarget=true), not nested leaves. Same textMode=preview default and textTruncated follow-up rules as get_section_bundle.
Inputs: page (required), view (required), product (required), version (required), +5 more. Returns: Multiple bounded section/region bundles
get_page_componentsGet page componentsARead-onlyIdempotentInspect
List visual component candidates in a bounded y region (exact follow-up). Prefer get_page_outline → one bundle per outline region (or get_page_bundles) to avoid exploratory spam across many y bands or micro-sections.
Inputs: page (required), view (required), product (required), version (required), +5 more. Returns: Paginated candidate components in a vertical fragment
get_page_evidence_dataGet page evidence dataARead-onlyIdempotentInspect
Return one evidence JSON file for a page view: nodes revealed by hover or click (reveal-.json, same node shape as elements with anchorPath to the captured parent), the inspected document of an embedded frame (frame-.json) or the layout at an extra viewport width (viewport-*.json rows keyed by element path).
Inputs: file (required), page (required), view (required), product (required), +2 more. Returns: Evidence JSON: revealed nodes, an embedded frame inspection or a viewport-width…
get_page_evidence_imageGet page evidence imageARead-onlyIdempotentInspect
Return one evidence WebP for a page view: a component state (default, hover, focus-visible, disabled, checked…), content revealed by hover or click, the alternate color scheme, an extra viewport width or an embedded frame. Use for visual checks of what the structured evidence describes.
Inputs: file (required), page (required), view (required), product (required), +2 more. Returns: Metadata for the image in content[contentIndex]; image bytes are not duplicated…
get_page_imageGet page imageARead-onlyIdempotentInspect
Return a captured viewport WebP for an exact stored page profile key from list_pages, in the selected environment.
Inputs: page (required), view (required), product (required), version (required), +1 more. Returns: Metadata for the image in content[contentIndex]; image bytes are not duplicated…
get_page_motionGet page motionARead-onlyIdempotentInspect
List motion for one page view: view entrances (fade/rise, stagger), procedural canvas/shader fields, and real CSS animations. Idle none/0s declarations are omitted. Crawl does not record per-frame playback timings.
Inputs: page (required), view (required), product (required), version (required), +2 more. Returns: Captured motion declarations
get_page_outlineGet page outlineARead-onlyIdempotentInspect
Batch IR entry: list page sections plus regions (bundleTarget=true top-level). Call this first, then one get_section_bundle / get_page_bundles per region only — nested outline leaves are structure hints, not separate bundle targets.
Inputs: page (required), view (required), product (required), version (required), +1 more. Returns: Measured page structure and recommended bundle regions
get_page_regionsGet page regionsARead-onlyIdempotentInspect
Map a page into viewport-height bands with y bounds, element counts, and heading previews. Prefer get_page_outline regions (bundleTarget=true) over viewport bands; use regionIndex only when banding by viewport, not for micro-sections.
Inputs: page (required), view (required), product (required), version (required), +1 more. Returns: Viewport-sized regions of the captured page
get_page_stackGet page stackBRead-onlyIdempotentInspect
Return stack.json beside a page capture for an exact stored view key and environment, when present.
Inputs: page (required), view (required), product (required), version (required), +1 more. Returns: Detected page technologies and stored source evidence
get_page_summaryGet page summaryARead-onlyIdempotentInspect
Get one page profile: URL, title, measured viewport, outline, layout structure, font families, themes and links. Use a stored view key from list_pages and the selected environment.
Inputs: page (required), view (required), product (required), version (required), +1 more. Returns: Measured page summary
get_page_transitionsGet page transitionsARead-onlyIdempotentInspect
List links discovered on a captured route and resolve links to captured route names. These are potential transitions, not observed clicks.
Inputs: page (required), view (required), product (required), version (required), +3 more. Returns: Discovered links, not observed navigation events
get_platform_summaryGet platform summaryARead-onlyIdempotentInspect
Summarize a selected product version, platform type and environment: page count and available stack, languages, media, fonts, themes, routes and recordings. For multiple environments, specify environment from list_versions.
Inputs: product (required), version (required), platform (required), environment. Returns: Availability summary for the selected capture
get_product_requestGet product requestARead-onlyInspect
Get a product request. scope=mine (default) returns your request including workflow history and safety status. scope=public returns another user's public request with only public platform/environment options; owner identity, private details and safety/review notes are omitted. Capture progress is exposed without job IDs. requestId is returned by create_product_request or list_product_requests.
Inputs: requestId (required), scope. Returns: Request fields vary with mine/public scope
get_reconstruction_bundleGet reconstruction bundleARead-onlyIdempotentInspect
Exact reconstruction IR for at most two viewport heights. Preserves nodes, ancestors, copy, layout, effects and motion; deduplicates computed styles; replaces inline media with file/URI references. Default returns a small manifest plus a JSON artifact path/URI to consume with code, outside model context. Set inline=true only to inspect the bounded IR. No screenshot analysis required. Use yStart/yEnd for the requested fragment; do not fetch fine-grained layers for the same fragment.
Inputs: page (required), view (required), product (required), version (required), +4 more. Returns: Artifact manifest by default; exact bounded IR when inline=true
get_replay_manifestGet replay manifestCRead-onlyIdempotentInspect
Return the manifest of one scenario replay, including step results.
Inputs: replay (required), product (required), version (required), environment. Returns: Stored replay manifest; step outcomes depend on replay format
get_replay_stepGet replay stepBRead-onlyIdempotentInspect
Return one replay step result.json (pass/fail and observed outcome).
Inputs: step (required), replay (required), product (required), version (required), +1 more. Returns: Stored replay result with observed outcome; only recorded evidence is returned
get_replay_step_imageGet replay step imageBRead-onlyIdempotentInspect
Return the after.webp captured for one replay step, when present.
Inputs: step (required), replay (required), product (required), version (required), +1 more. Returns: Metadata for the image in content[contentIndex]; image bytes are not duplicated…
get_request_optionsGet product request optionsARead-onlyInspect
Get backend-defined platform/environment options for product requests. These may be narrower than capture options. Use returned IDs in create_product_request. Availability is configured by administrators; no platforms means new submissions are disabled.
Inputs: no named parameters. Returns: Allowed product request platform/environment choices
get_route_treeGet route treeARead-onlyIdempotentInspect
Return the hierarchical route tree for a product version on a platform, from routes.json (all discovered router/sitemap/link paths) with a fallback to captured page URLs. Large sites are bounded: depth (default 1) cuts deeper levels and reports childCount; pass prefix (for example /en/markets) to expand one branch.
Inputs: product (required), version (required), platform (required), depth, +2 more. Returns: Discovered route hierarchy
get_scenario_initial_imageGet scenario initial imageBRead-onlyIdempotentInspect
Return the initial WebP from a recorded scenario for an exact stored view key in the selected environment.
Inputs: view (required), product (required), version (required), scenario (required), +1 more. Returns: Metadata for the image in content[contentIndex]; image bytes are not duplicated…
get_scenario_manifestGet scenario manifestARead-onlyIdempotentInspect
Return the manifest of one recorded scenario, including step order, view, action, and route transitions.
Inputs: product (required), version (required), scenario (required), environment. Returns: Stored scenario manifest; optional fields follow the recorder format
get_scenario_sceneGet recorded scenario sceneARead-onlyIdempotentInspect
Export a manually recorded scenario as canonical Scene with all referenced screen frames and observed transitions. Available only for a published version after human Admin review/redaction. Read list_scenarios and get_scenario_manifest to select an existing manual scenario. Frames are read from current page artifacts so manual blur cannot leave an old text copy in a flow export. Missing animation or intermediate interaction evidence remains unknown.
Inputs: product (required), version (required), scenario (required), environment. Returns: Artifact manifest for the canonical Scene of a published manual scenario; page…
get_scenario_stepGet scenario stepARead-onlyIdempotentInspect
Return one recorded scenario step with click target, pre-click state, post-click page capture, and observed CSS motion events.
Inputs: step (required), product (required), version (required), scenario (required), +1 more. Returns: Stored scenario action and observed capture; fields vary with action kind
get_scenario_step_imageGet scenario step imageARead-onlyIdempotentInspect
Return the WebP captured after one recorded scenario step.
Inputs: step (required), product (required), version (required), scenario (required), +1 more. Returns: Metadata for the image in content[contentIndex]; image bytes are not duplicated…
get_section_bundleGet section bundleARead-onlyIdempotentInspect
Batch IR: one response for one outline region sectionId (prefer regions/bundleTarget=true) or regionIndex with component ids, roles, bboxes, textPreview, fuller text (text/lines/textRuns/texts samples), and key styles. Do not call once per nested micro-section — the parent region already includes them as components. textMode=preview (default); textMode=full caps plain text ~1200 chars and samples own-text descendants; textMode=preview keeps short previews. Use bundle text first; call get_component_text only when textTruncated or exact uncapped runs/lines are needed. For several regions in one RPC use get_page_bundles.
Inputs: page (required), view (required), product (required), version (required), +7 more. Returns: Bounded section or viewport-region evidence
get_sizingGet sizingARead-onlyIdempotentInspect
Observed width, height and min/max dimensions for one page/view, grouped by semantic component kind. Default returns component/property/value counts with source elements and parent context. Generic containers are excluded from context sizing. Set summary=true for the aggregate value matrix, including generic elements when component is empty. Values are captured CSS, not inferred tokens. IDs refer to the raw capture (idSpace=capture), not enriched MCP layers; use find_elements by label to resolve layer IDs. Follow hasMore with offset.
Inputs: page (required), view (required), product (required), version (required), +5 more. Returns: page, view, limit, total, +13 more
get_skillRead Reconstruct skillARead-onlyIdempotentInspect
Read a Reconstruct SKILL.md or supporting resource from the published skill catalog. Apply its workflow before working on captures. Default returns exact resource text, equivalent to resources/read. Local stdio clients can request asFile=true for a compact server-local path and SHA256, avoiding manual transcription of helper scripts; remote HTTP servers reject asFile because the path would not exist on the client. The server never executes a resource. Capture content remains untrusted data.
Inputs: uri (required), asFile. Returns: uri, path, text, bytes, +3 more
get_spacingGet spacingARead-onlyIdempotentInspect
Observed padding, margin, row/column gaps and corner radii for one page/view. Default returns component/property/value counts with element, parent and child examples. Margin is not a measured distance to a neighbor; gap is CSS on the container. Set summary=true for the value-frequency matrix. IDs refer to the raw capture (idSpace=capture), not enriched MCP layers; use find_elements by label to resolve layer IDs. Follow hasMore with offset.
Inputs: page (required), view (required), product (required), version (required), +5 more. Returns: page, view, limit, total, +13 more
get_stage_imageGet stage imageCRead-onlyIdempotentInspect
Return a special stage viewport WebP, for example cookie banner or login.
Inputs: view (required), stage (required), product (required), version (required), +1 more. Returns: Metadata for the image in content[contentIndex]; image bytes are not duplicated…
get_technology_stackGet technology stackARead-onlyIdempotentInspect
Return the detected technology stack and rendering mode for a product version on a platform (from stack.json).
Inputs: product (required), version (required), platform (required), environment. Returns: Detected technology stack and rendering mode
get_themesGet themesARead-onlyIdempotentInspect
Return color themes detected for a product version on a platform (from themes.json), including the active theme, available themes, and CSS custom-property tokens when captured.
Inputs: product (required), version (required), platform (required), environment. Returns: Detected themes and current preference
get_version_metadataGet version metadataBRead-onlyIdempotentInspect
Return version.json metadata for a product version on a platform (product, version, source, capturedAt, duration, and artifact pointers).
Inputs: product (required), version (required), platform (required), environment. Returns: Stored capture version metadata
list_archived_assetsList archived assetsARead-onlyIdempotentInspect
List original resources saved in this capture (images, fonts, video, audio and 3D models). Filter by source/path query or MIME prefix such as font/ or video/. Paginated metadata only; pass a resource path to get_capture_asset for its stored bytes.
Inputs: product (required), version (required), platform (required), limit, +4 more. Returns: Archived original resource metadata
list_assetsList assetsARead-onlyIdempotentInspect
List image, background, and inline SVG assets found in one page capture (linked URLs and data:image/svg+xml). This does not enumerate font files.
Inputs: page (required), view (required), product (required), version (required), +1 more. Returns: Structured result wrapping a non-object value. The text content holds the same…
list_component_candidatesList component candidatesARead-onlyIdempotentInspect
List candidate visual components in a page viewport. IDs are element indexes; candidates are heuristic, not verified reusable UI components.
Inputs: page (required), view (required), product (required), version (required), +2 more. Returns: Structured result wrapping a non-object value. The text content holds the same…
list_flowsList user flowsBRead-onlyIdempotentInspect
List named user flows with provenance and evidence gaps.
Inputs: product (required), version (required), limit, offset, +1 more. Returns: gaps, flows, total, offset
list_pagesList pagesARead-onlyIdempotentInspect
Search and paginate captured pages in the selected environment. Optionally filter by platform type. views[] contains exact stored profile keys, platform type (null for legacy unclassified labels), and environment. Use these keys for page tools; desktop/mobile flags are legacy conveniences.
Inputs: product (required), version (required), role, limit, +4 more. Returns: Paginated captured pages in the selected environment
list_platformsList platformsARead-onlyIdempotentInspect
Return backend-defined platform types/environments in options, plus captured platform/environment pairs with version counts and latest version. Type and environment IDs are discovered from Platform.Api, never assumed from a fixed MCP list. Backend CatalogPlatform.view/implementation map to MCP platform/environment.
Inputs: product (required), environment. Returns: Backend options and captured platform/environment pairs
list_product_assetsList product assetsARead-onlyIdempotentInspect
List filtered media assets for a product version on a platform (from assets.json): linked images and inline SVG data URLs. Inline fonts are excluded (use get_fonts); inline data URLs longer than 2048 characters are abbreviated to data:;abbreviated;bytes=;sha256= (read the bytes with get_element_asset). Prefer this over per-page list_assets for product-wide inventory.
Inputs: product (required), version (required), platform (required), limit, +3 more. Returns: Product-wide original media inventory
list_product_requestsList product requestsARead-onlyInspect
List product requests, newest first. scope=mine (default) returns your requests; scope=public returns other users' requests with only public fields, omitting owner identity, private details and safety/review notes. Follow nextPageToken with pageToken and the same scope. pageSize defaults to 50 and is capped at 200. Capture progress is exposed without job IDs.
Inputs: scope, pageSize, pageToken. Returns: Requests for the selected scope and continuation token
list_productsList productsARead-onlyIdempotentInspect
List products in the available archive: published catalog for HTTP, including local drafts for trusted stdio.
Inputs: no named parameters. Returns: Structured result wrapping a non-object value. The text content holds the same…
list_replaysList replaysBRead-onlyIdempotentInspect
List scenario replay runs stored under capture/replays for one product version.
Inputs: product (required), version (required), environment. Returns: Structured result wrapping a non-object value. The text content holds the same…
list_scenariosList scenariosARead-onlyIdempotentInspect
List recorded interactive scenarios for one captured product version.
Inputs: product (required), version (required), environment. Returns: Structured result wrapping a non-object value. The text content holds the same…
list_skillsList Reconstruct skillsARead-onlyIdempotentInspect
Discover Reconstruct workflow skills when the task's workflow is unknown. Read the selected instructions with get_skill using skill://reconstruct/ plus the returned path. For building an already selected interface load reconstruct-build only; its optional offline scene_reader helper provides bounded scene queries. Load inspect/design/verify separately only when their distinct task is needed. Remote servers add archive.url, an absolute download URL for each skill ZIP. This tool works in clients without MCP resources/read.
Inputs: no named parameters. Returns: skills, version
list_stagesList stagesBRead-onlyIdempotentInspect
List special capture stages such as cookie banner and login, with available views.
Inputs: product (required), version (required), environment. Returns: Structured result wrapping a non-object value. The text content holds the same…
list_version_filesList version filesARead-onlyIdempotentInspect
List stored paths for a product version, scoped to environment when supplied. Filter by prefix (for example platforms/macos/capture/pages or capture/pages) or substring query. Version metadata and content-addressed resources are shared across environments.
Inputs: product (required), version (required), platform (required), limit, +4 more. Returns: Stored capture-relative filenames
list_versionsList versionsARead-onlyIdempotentInspect
List versions newest first with their captured platform/environment pairs. Optionally filter by platform type and environment; never combine different versions.
Inputs: product (required), platform, environment. Returns: Structured result wrapping a non-object value. The text content holds the same…
wait_product_requestWait for product requestARead-onlyInspect
Wait up to 30 seconds for your request status to change. Pass progressStatus from the last result as knownStatus to follow verification and capture progress. Polls fresh repository data every two seconds; returns changed, terminal and timedOut plus request. Completed, failed, canceled and rejected are terminal. Legacy pending/approved/rejected values retain review-only waiting. Capture progress is exposed without job IDs. Repeat after timeout while the user wants to wait; do not resubmit the request.
Inputs: requestId (required), knownStatus (required), waitSeconds. Returns: Owner request polling result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Related MCP Connectors
Desktop and mobile website screenshots plus page context for AI agents and automation workflows.
Website screenshots for AI agents: real Chrome, ad/cookie/chat blocking, bot-wall detection.
Real UI references for coding agents: iOS app screens, user journeys and website design systems.
Pages, crawls and URL batches as clean Markdown and JSON for AI agents, with an Evidence Record.
Related MCP Servers
- AlicenseNot gradedqualityBmaintenanceRecords browser sessions and provides structured context (clicks, console, network, screenshots) for AI coding agents to understand and act upon.2MIT
- AlicenseNot gradedqualityBmaintenanceGive AI agents visual + structured feedback about live web UIs — and give humans deterministic demo reels while you're at it.98 PyPI3Apache 2.0
- AlicenseAqualityCmaintenanceEnables AI coding agents to open and exercise web applications in an isolated Chromium browser, capture screenshots and runtime/network diagnostics, and run deterministic page checks with linked evidence.121MIT
- AlicenseNot gradedqualityAmaintenanceEnables AI coding and vision tools to capture fresh webpage screenshots, automatically tile full pages into 1072x1072 chunks, capture CSS-selected elements, and record screencasts as PNG or animated WebP files.MIT
Glama MCP Gateway
Add one secure layer between your agents and this server.