MCP developer bridge that forwards validated tool calls to a running Vespera Editor through localhost-only VAP v1 transport. Enables automation and testing of the Vespera game engine editor.
31 tools with consistent verb_noun naming (get_, list_, create_, set_, etc.) and detailed descriptions (avg 150-200 chars). All tools have input schemas with proper JSON Schema types and constraints (enums, minItems, required fields). However, output schemas are entirely undocumented, no tool specifies what fields it returns, forcing LLMs to infer structure. Parameter descriptions are absent from the schema definitions; only the C++ header provides semantic hints. Error handling and recovery guidance are not visible in the tool definitions. Security considerations (permission gates, audit trails) are not declared. Composition is strong, tools are single-responsibility and chain well via entity_id/asset_id references.
Add a built-in component by semantic component key.
Assign a project Material to an entity Mesh Renderer.
Create an empty, sprite, trigger or point-light entity.
Create a Lit or Unlit .slmat asset inside project Assets.
Create an undoable Cube, Plane, Cylinder or Sphere in the edit scene.
Delete one entity and its descendants through undoable editor history.
Duplicate one entity through the editor transaction path.
Output schemas completely undocumented. No tool specifies what fields it returns (e.g., vespera_get_entity returns entity data but structure is invisible). LLMs cannot plan downstream calls or extract required fields without guessing.
Parameter descriptions missing from JSON schemas. Only C++ header provides semantic hints (e.g., 'entity_id' has no description in schema, only in .hpp). LLMs cannot infer parameter meaning from names alone.
No error handling or recovery guidance visible. Tools do not document what errors can occur, whether they are retryable, or what the LLM should do next (e.g., if vespera_get_entity fails with 'entity not found', no guidance on recovery).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 61 | 2026-07-28+ | v2 |
Export/package the current project through Vespera's deterministic build closure and optionally launch the packaged runtime.
Read the MCP bridge version/tool count and live editor/runtime endpoints. Call this at the start of QA to detect stale MCP registrations.
Read the deterministic standalone-build asset closure and validation state.
Read recent typed editor command audit records, success state and transaction IDs.
Read one reflected built-in component property.
Read recent Vespera Editor Console entries for automated QA/failure diagnosis.
Read one entity including hierarchy, transforms and components.
Read the active Vespera project/scene, edit/play state and selection.
Inject/override a semantic named Input action into active Editor Play Mode without OS key synthesis.
Instantiate a project prefab by stable asset id or project-relative path.
List project assets with stable IDs, kinds and paths.
Read the editor extension API version and registered automation/extension commands.
List authored scene entities with IDs, hierarchy and component summary.
Control Vespera in-editor Play Mode.
Inject a deterministic runtime-UI pointer event into Editor Play Mode.
Redo the latest edit-scene transaction.
Refresh/reimport the project AssetCatalog.
Remove a removable built-in component by semantic component key.
Parent/reparent/unparent an entity while preserving world transform.
Select an authored entity by scene object ID.
Set one reflected built-in component property through undoable history. Values are strings or numeric arrays and are converted by reflected type.
Set position/rotation/scale for an entity as one undoable editor transaction.
Undo the latest edit-scene transaction.
Run Vespera scene validation and return error/warning counts.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) present. LLMs cannot distinguish safe reads from destructive writes without parsing descriptions. vespera_delete_entity should be marked destructive; vespera_undo/redo should be marked reversible.
No permission gates or scope declarations. Tools do not declare what permissions they require (e.g., 'write:scene', 'read:assets'). No audit trail or security context visible. Destructive tools like vespera_delete_entity lack confirmation patterns.