generated: '2026-09-07' method: searched status: published source: >- https://github.com/Facets-cloud/control-plane-openapi-mcp, https://github.com/Facets-cloud/control-plane-mcp-server, https://github.com/Facets-cloud/facets-module-mcp, https://github.com/Facets-cloud/praxis-cli, https://www.facets.cloud/docs/getting-started/build-with-an-ai-agent provider: Facets providerId: facets deployment: mode: local-stdio endpoint: null install: uvx control-plane-openapi-mcp package: https://pypi.org/project/control-plane-openapi-mcp/ auth: api-key verified: searched note: >- Facets ships THREE first-party MCP servers, all as PyPI packages a human installs and runs locally over stdio. There is no hosted, publicly addressable MCP endpoint an agent can POST to. facets-cp-mcp-server additionally supports a --transport streamable-http mode, but that server is one YOU run (default http://localhost:3000/mcp) - it is not a Facets-operated URL. Separately, Facets operates a server-side MCP gateway inside each customer's own control plane, reached through the `praxis mcp` CLI subcommand under the org's credentials; it is real agent infrastructure but it is per-tenant and is not a published https MCP endpoint, so it is recorded under gateway: below rather than as deployment.endpoint. auth is api-key because every server authenticates with FACETS_USERNAME plus a personal access token (HTTP Basic against the control plane). servers: - name: control-plane-openapi package: control-plane-openapi-mcp registry: pypi transport: stdio install: uvx control-plane-openapi-mcp repository: https://github.com/Facets-cloud/control-plane-openapi-mcp auth: style: env variables: [CONTROL_PLANE_URL, FACETS_USERNAME, FACETS_TOKEN, FACETS_PROFILE] credentials_file: ~/.facets/credentials description: >- Loads the live Control Plane OpenAPI from /v3/api-docs and exposes fuzzy search and detail lookup over its operations and schemas. The repository README states the catalog holds 566 operations of which 17 are deprecated, leaving 549 active, plus 500+ schemas. tools: - {name: FIRST_STEP_get_api_script_guide, description: Load the API script generation guide. Call first.} - {name: refresh_api_catalog, description: Re-fetch the latest OpenAPI spec from the control plane.} - {name: search_api_operations, description: Fuzzy search operations by id, summary, description, tag.} - {name: search_api_schemas, description: Search schemas by name and description.} - {name: load_api_operation_by_operationId, description: Load full operation detail by operationId.} - {name: load_api_operation_by_path_and_method, description: Load operation detail by path plus HTTP method.} - {name: load_api_schema_by_schemaName, description: Load a schema's properties, types and validation.} - {name: call_control_plane_api, description: Make an authenticated GET request to the Control Plane API.} - name: facets-control-plane package: facets-cp-mcp-server registry: pypi transport: stdio also_supports: streamable-http install: uvx facets-cp-mcp-server@latest repository: https://github.com/Facets-cloud/control-plane-mcp-server auth: style: env variables: [CONTROL_PLANE_URL, FACETS_USERNAME, FACETS_TOKEN, FACETS_PROFILE] description: >- Task-level tools over the control plane for project, variable, resource, environment and override management. Destructive operations require explicit confirmation and offer dry-run previews. tools: - {name: get_all_projects, category: project} - {name: get_project_details, category: project} - {name: use_project, category: project} - {name: refresh_current_project, category: project} - {name: get_secrets_and_vars, category: variable} - {name: get_variable_by_name, category: variable} - {name: create_variable, category: variable} - {name: update_variable, category: variable} - {name: delete_variable, category: variable} - {name: get_variable_environment_values, category: variable} - {name: update_variable_environment_value, category: variable} - {name: list_available_resources, category: resource} - {name: get_all_resources_by_project, category: resource} - {name: get_resource_by_project, category: resource} - {name: get_spec_for_resource, category: resource} - {name: get_module_inputs, category: resource} - {name: get_spec_for_module, category: resource} - {name: get_sample_for_module, category: resource} - {name: get_resource_schema_public, category: resource} - {name: add_resource, category: resource} - {name: update_resource, category: resource} - {name: delete_resource, category: resource} - {name: get_output_references, category: resource-config} - {name: explain_ui_annotation, category: resource-config} - {name: get_resource_output_tree, category: resource-config} - {name: get_resource_management_guide, category: resource-config} - {name: get_all_environments, category: environment} - {name: use_environment, category: environment} - {name: get_current_environment_details, category: environment} - {name: get_all_resources_by_environment, category: environment} - {name: get_resource_by_environment, category: environment} - {name: add_or_update_override_property, category: override} - {name: remove_override_property, category: override} - {name: replace_all_overrides, category: override} - {name: clear_all_overrides, category: override} - {name: preview_override_effect, category: override} - name: facets-module package: facets-module-mcp registry: pypi transport: stdio install: uvx facets-module-mcp repository: https://github.com/Facets-cloud/facets-module-mcp description: MCP server for authoring, validating and publishing Facets Terraform modules. tools: [] tools_note: Tool list not enumerated - the repository README documents a module-authoring workflow rather than a tool table. gateway: name: Praxis MCP gateway reachable_via: praxis mcp [ ] --json hosted: per-tenant, inside the customer's own control plane published_endpoint: null namespaces: - {name: k8s_cli, description: Read-only kubectl against connected clusters; credentials resolved server-side.} - {name: cloud_cli, description: Read-only aws / gcloud / az against the org's cloud integrations; mutating verbs blocked at the validator.} - {name: catalog_ops, description: List registered repos, search GitHub, register newly discovered repos.} note: >- Facets documents that "MCP tools run server-side under your organisation's managed credentials" so the developer's laptop needs no AWS, Kubernetes or Terraform credentials. The live namespace list is per-org - `praxis mcp` with no arguments returns it - so the three above are the ones the praxis-cli README names, not an exhaustive catalog. NOT recorded as deployment.endpoint because no https MCP URL is published. maintainers: - FN: Kin Lane email: kin@apievangelist.com