All servers / Developer / Superblocks MCP server
Superblocks MCP server
Superblocks MCP server
Tested by ConnectATDScore 86 out of 100. Tested 2026-11-19 on package version 2.0.179. Package: @superblocksteam/mcp-server · source
Score breakdown
| Runs | 18 / 20 | |
|---|---|---|
| Tool descriptions | 16 / 20 | |
| Package safety | 20 / 20 | |
| Maintenance | 20 / 20 | |
| Adoption | 12 / 20 |
What we saw
- Starts and answers the MCP handshake: yes (55.7 seconds including download)
- Tools offered: 74
- Last published: 2026-11-17
- Weekly downloads (npm): 1,380
- License: Superblocks Community Software License · Direct dependencies: 8
Want this server set up for your team?
ConnectATD builds AI automation and wires up MCP servers like this one. Tell us what you are trying to do.
Talk to ConnectATDTools and how they are described
Agents choose tools from these descriptions. Under each name you can see how many of its parameters are explained.
aggregate_audit_events13 of 14 parameters describedCompute aggregated metrics over v2 audit events (OCSF) for the current organization. Returns a single numeric value and optional grouped breakdown. Metrics: 'count' (total events), 'unique_actors' (distinct users), 'unique_resources' (distinct resources). Group by: 'activity_id', 'actor_user_uid', 'application_id', 'class_uid', 'day', 'hour', 'resource_type'. All the same OCSF filters from get_audit_events are available (class_uid, status, severity, actor, etc.). Example use cases: - Active end-users in a time range: metric=unique_actors - Active builders: metric=unique_actors, class_uid=6002 (APPLICATION_LIFECYCLE) - API executions by day: metric=count, class_uid=6003, status!=99, group_by=day IMPORTANT — API execution double-counting: API execution events (class_uid=6003) are logged as start/completion pairs. To count unique executions, filter to status != 99 (exclude IN_PROGRESS). Time range is capped at 366 days.
assign_builder_seat0 of 2 parameters describedAssign a builder seat to a user by user_id.
checkout_application2 of 6 parameters describedCheck out a Superblocks application to a caller-provided directory or a temporary directory and return the checkout path. Only Clark AI Apps (2.0) can be checked out. You can optionally provide commit_id to materialize a historical snapshot. Provide either app_url or application_id plus base_url.
create_folder0 of 2 parameters describedCreate a folder in the current organization.
create_integration7 of 9 parameters describedCreate a new integration. Call get_integration_config_schema first unless the missing-integration decision already includes configurationSchema. Pass name, plugin_id, and configuration. For credentialed integrations, include known non-secret fields and omit passwords, API tokens, and other secrets so the user can enter them in Superblocks. Do not send the integrations API envelope (id, created, configurations wrapper). To import a REST API from an OpenAPI spec, also pass openapi_spec (the spec file contents as JSON or YAML text) with file_name: the spec is validated, stored, and linked to the integration in the same call, and the result carries openapi_spec_ref. When the result carries integration_url, show it so the user can open the prefilled integration. An integration whose secrets you left blank cannot run a query yet, so during an import wait for the user to confirm they saved the credentials before calling edit_app with integration_id: Superblocks tests the APIs it wires, and every one of those tests fails on a blank secret. To reference the integration, include the returned mention unchanged in the next edit_app prompt.
delete_folder0 of 2 parameters describedDelete a folder by folder ID.
delete_group_org_role3 of 4 parameters describedRemove a group's organization-level role assignment by assignment record ID. This uses the dedicated org group role endpoint, not the generic RBAC assignment API. Use list_rbac_assignments to find the assignment_record_id. Requires group management privileges.
delete_integration0 of 2 parameters describedDelete an integration by ID. Integrations that back an App Database (managed_by = database_lifecycle in list_integrations) cannot be deleted here; retire the App Database instead.
deploy_application1 of 4 parameters describedDeploy a Superblocks application from its default branch. Supports either app_url or application_id plus base_url. Optionally pass commit_id to deploy a specific commit.
find_apps_by_integration2 of 4 parameters describedFind applications using a specific integration ID by querying server-side integration usage metadata.
find_apps_by_package5 of 9 parameters describedFind apps whose package usage index contains a package dependency. Supports filtering by deployed/non-deployed commits, declared or resolved version, and dependency source (root, direct, transitive).
get_app_database_pool_provisioning_run1 of 2 parameters describedGet one App Database pool provisioning run by the provisioning_id reported by list_app_database_pools or get_app_databases_overview: its status (provisioning, active, or failed), who started it and when, the physical instance it registered once active, and the error if it failed. Requires the app_databases:manage permission (Admin or Owner by default) and App Databases to be enabled for the organization.
get_app_databases_overview0 of 1 parameter describedOrg-wide App Databases summary: the total number of database records (database_record_count, which includes retired databases; use list_app_databases to enumerate live ones), per data tag how many applications and instances it has and whether a lifecycle agent serves it, every physical database instance with its capacity and per-state database counts, the physical pools each agent advertises, and which pools and data tags can be provisioned right now. Use list_app_databases for the individual databases. Requires the app_databases:manage permission (Admin or Owner by default) and App Databases to be enabled for the organization.
get_application_access0 of 4 parameters describedList all users and groups that have access to a Superblocks application, including their roles and permissions. Provide either app_url or application_id plus base_url.
get_application_commit_history2 of 6 parameters describedFetch a paginated, reverse-chronological application commit history, including autosaves, commit metadata, committer details, and DBFS directory hashes when available. Provide either app_url or application_id plus base_url.
get_application_database_compatibility0 of 3 parameters describedReport whether one application's App Database can be served on every data tag the application can run under: whether it declares an App Database at all (has_database), the overall state (overall_state: compatible, temporarily_unavailable, or incompatible), whether someone has to change the application (needs_attention), and per data tag (data_tags) which lifecycle operations no connected agent supports. For an application without an App Database, needs_attention is always false (there is nothing to fix) and overall_state says whether a database could be added: the server refuses to create one unless it is compatible, and data_tags show which tags fall short. The verdict fields match the app_database block of get_application_summary. Use it to explain why publishing or editing an application is blocked on its database. Readable by anyone who can view the application. Supports either app_url or application_id plus base_url.
get_application_deployment_history2 of 6 parameters describedFetch the deployment timeline for a Superblocks application. Returns deploy and undeploy events in reverse chronological order, including who performed each action and when. Undeploy entries have a null commitId. Supports either app_url or application_id plus base_url.
get_application_policy_run_findings3 of 8 parameters describedGet the app-scoped, paginated findings for one policy agent run on a specific application (the publish tab's findings table).
get_application_publish_pipeline1 of 3 parameters describedGet the status-only publish plan for one application commit: the ordered steps (build, policy checks, lifecycle, deploy) that publishing would run, without starting anything.
get_application_publish_readiness1 of 3 parameters describedGet the policy-gate publish readiness for one application commit: which policy checks are required, their live status, and whether the commit is ready to publish.
get_application_structure0 of 5 parameters describedFetch a Superblocks application structure summary. For Clark AI Apps this returns a bounded checkout tree and temp path. For app behavior questions, debugging questions, and investigation of bug reports, if edit_app is available, ask Superblocks through edit_app in Plan mode and pass the user's words verbatim instead of inspecting source here. Otherwise, use this tool to inspect the application. When calling edit_app, use mode "BUILD" only when the user explicitly asks to skip planning. Provide either app_url or application_id plus base_url.
get_application_summary0 of 4 parameters describedFetch a Superblocks application summary, including the owning organization ID, app mode, deployed commit ID when available, resolved environment, and App Database status (has_database, overall_state, needs_attention; for an application without a database needs_attention is always false and overall_state says whether one could be added, since the server refuses to create one unless it is compatible; null with app_databases_availability explaining why when the organization has App Databases disabled, the server predates them, or the App Databases read failed; the summary itself never fails because of App Databases). Provide either app_url or application_id plus base_url.
get_app_sbom1 of 3 parameters describedFetch the software bill of materials (SBOM) for a Superblocks application. Returns a 'sbom' field containing an unmodified CycloneDX 1.5 JSON document. Defaults to the app's currently-deployed commit; pass commit_id to fetch a specific commit's SBOM instead.
get_audit_events12 of 14 parameters describedFetch v2 audit events (OCSF) for the current organization in a time range, with cursor pagination and OCSF filters. Use list_audit_event_types to discover all available class_uid, resource_type, and operation values for filtering. IMPORTANT — API execution double-counting: API execution events (class_uid=6003) are logged as start/completion pairs. Each execution produces two events with the same timestamp: a start event (status_id=99, no duration) and a completion event (status_id=1 for success or status_id=2 for failure, includes duration in ms). To count unique executions, filter to status_id != 99 (i.e. exclude IN_PROGRESS) or divide the raw class_uid=6003 count by 2.
get_auth_status0 of 2 parameters describedCheck whether the current user is authenticated for a Superblocks environment derived from either an app_url or a base_url.
get_billing_plan_summary0 of 1 parameter describedGet the current billing plan summary including credit allocation, usage, billing cycle dates, and deployed application counts. Requires billing:manage permission.
get_billing_usage_daily2 of 3 parameters describedGet daily AI credit usage for a date range. Returns one row per day with the total credits consumed. Requires billing:manage permission.
get_billing_usage_records6 of 7 parameters describedGet AI credit billing usage for a date range, rolled up per checkpoint (an accepted AI change). Each entry has checkpoint_id, checkpoint_label (the checkpoint's commit message; null when the commit had no message), application, credits_used, and overage splits. Usage not attributable to any checkpoint (history from before checkpoint tracking shipped, never-accepted prompts, and agent or utility usage) is aggregated into a single trailing entry with checkpoint_id null — expected, not missing data; older date ranges fall mostly in that bucket. Because that entry sorts last, it may sit past the requested page; no_checkpoint_credits always reports its credit total for the whole filtered range, so compare it against total_credits before summarizing a partial page. dashboard_credits is the part of that usage spent in the dashboard Clark chat, outside any app (null when the server does not report it). checkpoint_data_available is false when the server predates checkpoint rollups (no_checkpoint_credits is then null — unknowable, not zero). App filters (application_id, usage_type) scope every surface, including checkpoints. Pagination (limit/offset) applies only to the checkpoints list; by_application, dollar_ledger_records, and top_users are always full summaries. by_application credits are an independent server-side per-application total and are not expected to equal the sum of that application's checkpoints. Requires billing:manage permission.
get_chat_messages1 of 7 parameters describedFetch persisted Builder chat messages for a Superblocks application. By default this returns the editor transcript from the active session (including user and assistant/status messages) so the message stream matches what the Builder shows. If user_id is provided, the response is filtered to user-authored messages for that sender. Returns 50 messages per page unless limit is set.
get_current_user0 of 1 parameter describedGet the currently authenticated Superblocks user: their identity, organizations, and permissions. Use list_organization_agents for agent details.
get_dependency_scan_summary0 of 1 parameter describedGet the org-wide dependency scan summary cards (Security Center CVE fleet view): overall severity counts and how many published apps have active vulnerabilities.
get_integration0 of 2 parameters describedGet a Superblocks integration by ID, including configuration metadata when available.
get_integration_access0 of 2 parameters describedList all users and groups that have access to a Superblocks integration, including their roles and permissions. Provide integration_id plus base_url.
get_integration_config_schema1 of 2 parameters describedReturn the JSON schema for a plugin's configuration object. Call this before create_integration. Pass the returned configuration fields as configuration; do not send API envelope fields such as id, created, integrationId, or bindingFields.
get_job_access0 of 2 parameters describedList users and groups that have access to a scheduled job, including role and permission details.
get_knowledge1 of 2 parameters describedGet a single knowledge entry (fact) by ID, including its full content. Knowledge entries are persistent notes and instructions that Clark uses when building apps.
get_review_policy0 of 2 parameters describedGet the full configuration of a single policy agent / review policy, including its prompt, mode (advisory or blocking), scope, stage, and domain.
get_review_policy_run_report3 of 7 parameters describedGet the full report for a single policy agent run: aggregate blocking/non-blocking counts and severity breakdown, plus the paginated list of individual findings for that run.
get_review_policy_summary1 of 2 parameters describedGet org-wide policy agent summary stats over a time window (default 24h): active policy count, total runs, findings count, and pass rate.
get_user_permissions0 of 3 parameters describedGet all resources a user can access and their role on each, including assignments inherited through groups. Returns RBAC assignments across applications, integrations, workflows, and jobs. Useful for deprovisioning audits, compliance reviews, and access certification.
get_workflow_access0 of 2 parameters describedList users and groups that have access to a workflow, including role and permission details.
grant_resource_access0 of 2 parameters describedCreate RBAC assignments for a resource (roles or direct permissions). Organization scope is always the authenticated user's current organization (matches server enforcement).
list_app_database_pools0 of 1 parameter describedList every App Database physical pool in the organization, including pools still provisioning and pools whose last provisioning run failed. A pool is one (agent name, engine) pair owned by a lifecycle agent; each entry carries its registered physical instances with endpoint, region, status, and capacity, plus the latest provisioning run (status, who started it, when, and the error if it failed). Requires the app_databases:manage permission (Admin or Owner by default) and App Databases to be enabled for the organization.
list_app_databases3 of 4 parameters describedList the App Databases in the organization, one row per database with the application it belongs to, its data tag, lifecycle and migration state, the physical instance hosting it, and its connection host/port/database. Retired databases (torn down, retired_at set) are left out unless include_retired is true; the tool reads further server pages as needed to fill a page of active rows. offset and next_offset count server records (retired included), and total_count is the number of database records the server holds, retired included. Requires the app_databases:manage permission (Admin or Owner by default) and App Databases to be enabled for the organization.
list_applications11 of 12 parameters describedList all Superblocks applications the current user can access. Supports filtering by creation/update date range, name substring, creator email, deploy status, application mode, and whether the app has an App Database. Each row reports has_app_database; app_databases_availability says why it is null when it is (App Databases disabled for the organization, an older server without them, an App Databases read error, or an empty page). The application list itself never fails because of App Databases.
list_audit_event_typesno parametersReturns the full catalog of audit event types emitted by Superblocks. Use this to discover valid class_uid, resource_type, and operation values for filtering with get_audit_events. No API call is made — this is a static reference.
list_billing_usage_limits2 of 3 parameters describedList configured AI usage limits for the current organization, including per-user/default user limits and the enterprise org-wide overage limit when available. Requires billing:manage permission.
list_builder_seats0 of 1 parameter describedList builder seats for the current organization.
list_cve_exceptions0 of 1 parameter describedList the Superblocks-managed CVE exceptions (the Security Center Exceptions tab): advisories accepted for packages a customer cannot change. Each entry carries the affected package, every advisory id for the vulnerability, the acceptance category ('false_positive', 'version_conflict', or 'deferred'), the customer-visible rationale, the acceptance date, the review expiration date, and the expected patch date for a deferred entry. Exceptions carry no severity because they are accepted findings, not active ones. Read-only: exceptions are maintained by Superblocks and are the same for every organization.
list_dependency_scan_findings7 of 11 parameters describedList per-application CVE / dependency vulnerability findings across the org's fleet (Security Center dependency scan view). Filter by application, one or more severities, a free-text search term, or publish/update date window. Defaults to hiding apps with no active vulnerabilities.
list_distinct_packages_in_org4 of 7 parameters describedList the distinct set of packages declared across every application in the current org's package usage index. Inverse of find_apps_by_package: returns one row per package_name with the set of declared versions, the count of apps using it, the deployed subset, and a bounded sample of app IDs (≤10). Use this when reconciling the org's package footprint against an external registry or CVE feed.
list_folders0 of 1 parameter describedList folders in the current organization, including IDs and metadata.
list_integrations6 of 7 parameters describedList all integrations (data sources, APIs, secret stores) configured in the current organization. Optionally filter by plugin_id (the integration type, such as anthropic or postgres), kind, creator email, or name substring. This returns integrations someone has already configured, so an empty result means none exists yet, not that the type is unsupported. To reference a selected integration, include its mention unchanged in the next edit_app prompt. Each row reports managed_by, application_id, and user_editable; integrations managed by database_lifecycle back an App Database and cannot be updated or deleted through the integrations API.
list_knowledge9 of 10 parameters describedList knowledge entries (facts) for the current organization. Knowledge entries are persistent notes and instructions that Clark uses when building apps. Supports filtering by scope, access type, enabled status, and text search.
list_mcp_client_requests1 of 2 parameters describedList the AI clients that signed in to the organization's Superblocks MCP server, with their status: pending (waiting for an admin), approved (can use Superblocks tools) or rejected. Each entry has the client ID (for most clients the URL of its metadata document), its domain, who asked first, and when it was first and last seen. Judge a client by its domain: anyone can publish a client named after a well-known product.
list_organization_agents2 of 3 parameters describedList non-deactivated agents for the authenticated user's current organization. Organization scope matches auth (same as other MCP tools); organization_id is optional and only needed if you must target a specific org UUID for URL compatibility.
list_packages_by_app7 of 11 parameters describedList package dependencies indexed for a Superblocks application. Mirrors `find_apps_by_package` against the same usage index but pivots on the app: given an application_id, returns the packages declared in package.json (and optionally resolved from the lockfile) for one or more indexed commits. Supports filtering by deployed/non-deployed commits, specific commit, declared or resolved version, package name, and dependency source (root, direct, transitive).
list_rbac_assignments1 of 3 parameters describedList RBAC assignments for a resource by resource_type and resource_id (for example apps, integrations, workflows, jobs).
list_review_policies0 of 1 parameter describedList every policy agent / review policy configured for the org (Security Center Policies view): name, policy type, status, mode (advisory or blocking), scope, stage, domain, and the last run summary for each.
list_review_policy_runs7 of 13 parameters describedList policy agent run history, either org-wide or scoped to a single policy via policy_id. Each run reports its status, decision, findings count, and the triggering app/commit/user.
list_roles2 of 3 parameters describedList available RBAC roles for an organization, including their UUIDs. Use these role UUIDs as assignment_id when calling grant_resource_access. Optionally filter by role type (apps, integrations, org, jobs, workflows). Organization scope defaults to the authenticated user's current organization; pass organization_id only when targeting a different org you belong to.
list_users3 of 5 parameters describedList users in a Superblocks organization with their status, type, and org-level role assignments, one page at a time. Pass search (part of an email or name) to find a specific user, for example to get the user_id other user tools need, instead of paging through everyone. Follow next_offset while has_more is true to read further pages. Useful for access audits, user inventories, and governance workflows.
list_users_by_role3 of 5 parameters describedList users in a Superblocks organization filtered by org role, one page at a time. role_name is an org role name (e.g. Admin, Developer, End-User), matched case-insensitively, or an org role UUID from list_roles. Follow next_offset while has_more is true to read further pages. To find one user by email or name, use list_users with search instead. Useful for access audits and compliance reviews.
lookup_package_vulnerabilities1 of 2 parameters describedCheck a batch of npm {name, version} pairs against the org-agnostic package vulnerability index before an install. Returns per-package severity, covering advisories, whether the package carries known malware, and a non-binding remediation suggestion ('pass', 'remediatable', or 'block').
revoke_resource_access1 of 2 parameters describedDelete a single RBAC assignment by assignment record ID.
set_mcp_client_approval2 of 3 parameters describedReject an AI client's request to use the organization's Superblocks tools, or revoke an approved client, by setting status 'rejected'. Approving (status 'approved') is always refused here, so an AI client can never approve another one without the user: an admin approves clients in the Superblocks organization settings.
unassign_builder_seat0 of 2 parameters describedUnassign a builder seat for a user by user_id.
undeploy_application1 of 4 parameters describedTake a deployed Superblocks application offline. The app becomes inaccessible to all users until redeployed. Supports either app_url or application_id plus base_url. Optionally pass a custom message for the audit trail.
update_folder0 of 3 parameters describedUpdate a folder name by folder ID.
update_group_org_role3 of 4 parameters describedChange a group's organization-level role (e.g. Admin, Developer, End-User). This uses the dedicated org group role endpoint, not the generic RBAC assignment API. Use list_roles with type='org' to find available role IDs. Requires group management privileges.
update_org_billing_usage_limit2 of 3 parameters describedSet the enterprise org-wide overage AI usage limit. Provide either credit_limit or amount_limit_cents. Requires billing:manage permission and an enterprise plan.
update_resource_access3 of 4 parameters describedUpdate a single RBAC assignment by assignment record ID.
update_user_billing_usage_limit2 of 4 parameters describedSet a per-user AI usage limit override, or omit user_id to set the default per-user usage limit for the organization. Provide either credit_limit or amount_limit_cents. Requires billing:manage permission.
update_user_org_role2 of 4 parameters describedChange a user's organization-level role (e.g. Admin, Developer, End-User). This uses the dedicated org user role endpoint, not the generic RBAC assignment API. Use list_roles with type='org' to find available role IDs. Requires org admin privileges.
Tested in a sandbox with an empty environment: no API keys, a throwaway folder, nothing from a real account. A passing test shows the server runs and responds. It is not a security audit.