@weppy/roblox-mcp
Installed, started with no arguments, no environment and no input, and listed its tools. Run 2026-11-16.
Needs to start: nothing
Contacts while installing and starting: api.ipify.org, weppy-operations-server.hope841026.workers.dev
What it contacted
api.ipify.orgweppy-operations-server.hope841026.workers.dev
Hosts it looked up or connected to while installing and starting with no input, watched from outside its sandbox. The npm registry is left out. Says nothing about what its tools contact when called.
Starting it
npx -y -p @weppy/roblox-mcp@2.17.22 roblox-mcp
Installing a package runs its code. Use a sandbox for anything you have not read.
What its tools tell an agent
In the tool descriptions, nothing that talks about the user, other tools or private files, or uses hidden characters.
1 tool with commands in capitals or very long text
- Very long description in
manage_ui:3000 characters or more
Common and usually harmless: it is how many makers stress how a tool should be used.
Found by simple pattern rules in each tool's description (up to 3,000 characters), quoted as written. Parameter descriptions are not scanned. A clean result is not a safety review.
Tools
query_instances action, path, sessionDebugId, siblingIndex, childName, descendantName, className, recursive, maxDepth, timeout, root, query, propertyName, propertyValue, tag, includeServices, includeSubclasses, caseSensitive, maxResults, depth, placeId, clientId, targetAlias
Query Roblox instances: get, children, find child/descendant, wait for child, class info, search by name/class. [PRO] file_tree, project_structure, descendants, ancestors, search by property/tag.
mutate_instances action, contextId, contextSummary, replayMetadata, className, parent, name, properties, path, sessionDebugId, siblingIndex, sourcePath, targetParent, newName, newParent, position, cframe, offset, tree, instances, paths, count, templatePaths, region, seed, maxSlope, avoidWater, alignToNormal, scaleJitter, minSpacing, parentName, placeId, clientId, targetAlias
Create, delete, clone, move, rename, or pivot instances. [PRO] create_tree, mass_create, mass_delete, mass_duplicate, smart_duplicate, scatter.
manage_properties action, contextId, contextSummary, replayMetadata, path, property, value, properties, attribute, tag, tagName, root, maxResults, expression, variables, operation, amount, paths, parentPath, modifications, propertyName, propertyValue, filter, recursive, includeReadOnly, placeId, clientId, targetAlias
Get/set properties, attributes, and tags on instances. [PRO] set_calculated, set_relative, mass_set, mass_get, modify_children.
manage_scripts action, contextId, contextSummary, replayMetadata, path, source, validate, scriptType, parent, name, startLine, endLine, newLines, lines, afterLine, content, pattern, replacement, caseSensitive, wholeWord, usePattern, dryRun, maxResults, placeId, clientId, targetAlias
Manage script source code: read, write, create, delete, edit lines, search, and validate Luau syntax. [PRO] replace across scripts.
manage_ui action, designId, expectedRevision, saveRequestId, limit, offset, designRef, design, editGuard, contextId, contextSummary, replayMetadata, brief, projectContext, agentCapabilities, briefId, runtimeEvidenceRef, observationMode, viewportMode, profileIds, tree, targetPath, changes, visualReview, placeId, clientId, targetAlias
Create, refine, inspect and verify editable Roblox in-game UI. Read `weppy://ui-studio/guide` for the self-contained workflow and supported evidence; `weppy://ui-studio/tokens` offers optional visual examples and `weppy://ui-studio/functional-rules` defines advisory checks and official platform guidance. Distinguish first creation, quality improvement, narrow editing and review-only, then choose presentation independently from screen job. Compose around player decisions: meaningful values, objects, states and actions should lead the hierarchy. A missing game concept does not select an app skin or a default genre. In improvement, preserve behavior/data and explicit layout choices, but allow justified changes to size and allocation rather than adding a skin to a faulty hierarchy. Use native UI for editable content and simple surfaces; reuse reviewed parts or generate authorized parts where useful, checking actual small-size readability, alpha boundaries and crop/slice assembly. These are design recommendations, not style quotas or new prerequisites. `design_brief` optionally inspects Studio context and returns candidate brief patches; create_tree/update may omit briefId. Do not dump enum lists or require every brief field. An explicitly supplied empty, unknown, expired or unreadable briefId fails before mutation: resolve it or explicitly omit it instead of silently discarding it. Keep accepted brief_id, design_contract and advisory Layout Plan context through related calls; explicit user/projectContext evidence wins. Existing-UI review cannot infer runtime absence from empty StarterGui: follow runtime_review_required with a structured Play Client PlayerGui snapshot and saved runtimeEvidenceRef. Saved Designs are Place-scoped appearance evidence, not complete screens or verified readiness. Optional editGuard protects property-only identity/value expectations; inspect actual edit_outcome and changed_properties. agentCapabilities declares provider-neutral image generation capability, review and local output, not authorization. Always present an early asset recommendation for new UI or visual improvement, including the role, reason and source or production route. Let the user decide whether to apply it, honor an existing choice or delegation, and use the available evidence even if there are no candidates or no brief. Proceed within existing authorization. manage_assets.search requires approved intent. External screenshot search, concept/part generation and Roblox upload follow their actual user-authorized scopes; reuse granted permission and keep a supported assetless fallback. Preview and calling-agent visualReview (`recompose`, `refine`, `approved`) are optional. Mutation success, visual judgment, functional_safe and readiness are distinct; disclose skipped coverage, with responsive evidence required only when responsive_requirement says required. Keep observed platform_ui, official platform_guidance and AI interpretation separate. Review-only does n
20 more tools
manage_lighting action, contextId, contextSummary, replayMetadata, properties, createIfMissing, time, clockTime, preset, overrides, placeId, clientId, targetAlias
[PRO] Configure environment: lighting, atmosphere, sky, terrain properties, time of day.
manage_selection action, paths, maxAge, maxDepth, includeSource, includeProperties, includeChildren, includeAncestors, placeId, clientId, targetAlias
Get, set, or clear selection. [PRO] context, details, add/remove items, watch changes.
manage_camera action, path, position, distance, duration, offset, lookAt, respectAutoFocusSetting, placeId, clientId, targetAlias
Camera operations: get info, focus on instance/position, suggest view, capture Edit-mode viewport screenshot (Edit mode only; not usable during playtest).
manage_tween action, contextId, contextSummary, replayMetadata, path, tweenInfo, properties, tweenId, placeId, clientId, targetAlias
[PRO] Tween service: create, play, pause, cancel tweens for smooth animations.
manage_audio action, contextId, contextSummary, replayMetadata, path, soundId, volume, looped, playbackSpeed, listenerType, listenerPath, placeId, clientId, targetAlias
[PRO] Audio management: play, stop, pause, resume sounds. Set audio listener.
manage_animation action, contextId, contextSummary, replayMetadata, path, animationId, trackId, speed, fadeTime, weight, priority, placeId, clientId, targetAlias
[PRO] Animation: load, play, stop animations. Get animation tracks from humanoid/controller.
manage_physics action, contextId, contextSummary, replayMetadata, groupName, group1, group2, collidable, placeId, clientId, targetAlias
[PRO] Physics collision groups: register, set collidable between groups, list groups.
manage_effects action, contextId, contextSummary, replayMetadata, path, count, enabled, placeId, clientId, targetAlias
[PRO] Particle effects: emit particles, clear all particles, toggle effect enabled state.
manage_terrain action, contextId, contextSummary, replayMetadata, material, center, position, cframe, size, radius, height, region, min, max, sourceMaterial, targetMaterial, color, resolution, materials, occupancy, baseHeight, amplitude, frequency, seed, layers, intensity, preset, octaves, persistence, waterLevel, edgeBlend, materialPalette, placeId, clientId, targetAlias
[PRO] Terrain operations: fill shapes, clear regions, replace materials, manage colors, read/write voxels, generate procedural terrain, smooth terrain.
spatial_query action, origin, direction, maxDistance, position, size, filterType, filterList, filterInstances, ignoreWater, rays, center, area, searchArea, minSize, maxSlope, resolution, gridSize, path, paths, tolerance, gridSnap, axes, ignorePaths, offset, rotation, checkGround, spawnSize, minSpacing, preferOutdoor, count, characterHeight, maxStepHeight, rootPath, includeModels, maxResults, padding, placeId, clientId, targetAlias
[PRO] Spatial queries: raycast, find ground, check placement, multi-raycast, scan area, find flat areas, find spawn positions, analyze walkable, spatial map, find empty space, get bounds, snap to grid, check collision.
manage_assets action, contextId, contextSummary, replayMetadata, assetId, parent, name, position, query, maxResults, category, sortType, format, includeProperties, includeChildren, maxDepth, scope, placeId, assetLibraryAssetId, targetParent, sourcePath, displayName, description, assetType, thumbnailMode, uploadMode, uploadTemporaryAssets, creatorType, creatorId, prompt, schemaMode, predefinedSchema, schemaGroups, size, maxTriangles, generateTextures, review, exportToLibrary, expectedUse, expectedGroups, maxDescendants, readiness, anchorMode, clientId, targetAlias
[PRO] Asset management: insert models by ID, get asset info, search creator store, insert free models/packages, export selection JSON, generate/review Roblox models, round-trip .rbxm files through Asset Library, and generate local thumbnails.
manage_open_cloud_assets action, contextId, contextSummary, replayMetadata, assetId, operationId, filePath, category, scope, placeId, displayName, description, creatorType, creatorId, expectedPrice, readMask, updateMask, waitForOperation, operationPollTimeoutMs, operationPollIntervalMs
[PRO] Roblox Open Cloud asset upload: preflight diagnostics, credential status, category capabilities, Asset Library-backed upload, existing asset linking, updates, metadata reads, and operation polling. Does not expose delete/archive/restore actions.
manage_sync action, placeId, instancePath, content, query
[PRO] Project sync management: status, history, effective policy, and synced file access.
workspace_state action, includeMetadata, limit, includeCameraInfo, includeSelectionBounds, placeId, clientId, targetAlias
[PRO] Workspace state: full sync, snapshot, recent changes, viewport info, clear history, metadata, scripts, selection info, clear cache.
manage_logs action, level, limit, pattern, since, sinceSeq, placeId, clientId, targetAlias
Output logs: get filtered logs, poll incrementally with sinceSeq cursor, clear buffer, get recent errors.
manage_settings action, enabled
WEPPY settings management. Read or change whether Dashboard opens automatically when the MCP Server starts.
system_info action, message, placeId, clientId, targetAlias
System info: ping, connection status, usage tier, and read-only Studio preflight diagnostics. [PRO] place info and services list.
manage_studio action, contextId, contextSummary, replayMetadata, enabled, mode, script, test_name, timeout, testProfile, restoreAfterTest, scenario, idempotencyKey, sessionId, controlProfileId, controlProfile, profileOperation, sourceProfileId, sourceRevision, sourcePlaceId, sourcePlaceName, destinationProfileId, destinationProfileName, candidateSource, candidateTarget, candidatePath, maxResults, adapterOperation, adapterId, adapterRegistrationId, adapterRegistrationLabel, adapterRegistrationGroup, adapterRegistrationKind, adapterValueType, adapterModulePath, adapterReadMethod, adapterWriteMethod, adapterMethodStyle, adapterStudioOnly, adapterConstraints, adapterEnumValues, productionExclusionStrategy, productionExclusionReference, planToken, expectedSourceHash, leaseId, expectedRevision, controlValues, actionId, actionArguments, expectedProfileRevision, requestId, stepCursor, stepLimit, locale, placeId, clientId, targetAlias
Prepare Playtest values on Basic, and control Roblox Studio playtest lifecycle, automated test runs, runtime values, and editor view/rendering QA workflows on Pro.
batch_execute commands, stopOnError, placeId, clientId, targetAlias
[PRO] Execute multiple commands in a single batch. Each command is an object with "tool" name and "args". Commands execute sequentially; optionally continue on error.
execute_luau contextId, contextSummary, replayMetadata, source, placeId, clientId, targetAlias
[PRO] Execute arbitrary Luau code in Roblox Studio sandbox. Blocked services: HttpService, DataStoreService, MessagingService. Cannot access CoreGui/CorePackages.
Descriptions are the package's own words, cut at 3,000 characters. Treat them as data.
Files it wrote
/tmp/ .weppy-data/ weppy-project-sync/ .gitignore /tmp/ .weppy-data/ telemetry-first-seen.json /tmp/ .weppy-data/ runtime/ bridge-startup.lock /tmp/ .weppy-data/ runtime/ bridge.log /tmp/ .weppy-data/ runtime/ bridge-state.json /tmp/ .weppy-data/ device-id.json /tmp/ Documents/ Roblox/ Plugins/ WeppyRobloxMCP.rbxm
Paths inside its sandbox, where its home folder is /tmp and its own folder cannot be written to.
Install facts
| License | SEE LICENSE IN LICENSE |
|---|---|
| Needs | node >=22.0.0 |
| Source | github.com/hope1026/weppy-roblox-mcp |
| Install scripts | None declared by it or anything installed with it. |
| Tarball | sha512-z/thF/iEIe5ZaivUGpyG2Q287efdbXWCVjZghKFLoL7LHlSBnLt0bGYQ3Yv/go3p2iwx+TR6pGoDi0i+jLaAGQ== |
Check this report yourself
This site did not write the report. A public job on a GitHub-hosted machine did: the run, from this commit of the open-source runner. GitHub signed a token that names that job and the hash of the report, so any edit since would show.
Report as JSONExact text and tokenHow to verify it
Show it on your README
[](https://bragproof.smaverk.com/r/%40weppy/roblox-mcp)
New version out? Ask for a fresh report.