@lightsage/mcp 0.3.0 → 0.3.1

package/esm/tool-names.js

@@ -110,19 +110,19 @@ export const toolNames = [
     },
     {
         "name": "evals-v2-evals-list",
-        "description": "List Evals\n\nReturns all active evals including workflow, MCP, CLI, and skill evals owned by the organization."
+        "description": "List Evals\n\nReturns all active eval definitions owned by the organization."
     },
     {
         "name": "evals-v2-evals-create",
-        "description": "Create Eval\n\nCreates an eval with a task prompt, subject, and plain-text success criterion in `judge`. The eval can then be run through eval runs."
+        "description": "Create Eval\n\nCreates a reusable simulation definition with one or more task prompts and judges. Optional agents, persona, CLI installations, MCP servers, repeat count, and tags are saved with the eval. Creating the eval does not start a run and does not accept secret environment values."
     },
     {
         "name": "evals-v2-evals-retrieve",
-        "description": "Retrieve Eval\n\nReturns one eval and its judge configuration."
+        "description": "Retrieve Eval\n\nReturns one reusable eval simulation definition."
     },
     {
         "name": "evals-v2-evals-update",
-        "description": "Update Eval\n\nUpdates an eval's task prompt, subject, plain-text `judge`, or name. Omitted fields remain unchanged."
+        "description": "Update Eval\n\nUpdates a reusable eval simulation definition. Omitted fields remain unchanged. Empty `agent`, `clis`, `mcps`, or `tags` arrays clear those collections, and `persona: null` clears the saved persona."
     },
     {
         "name": "evals-v2-evals-delete",
@@ -150,7 +150,7 @@ export const toolNames = [
     },
     {
         "name": "evals-run",
-        "description": "Start an eval run\n\nQueues a durable eval job and returns immediately with a polling ID. Poll `GET /v2/eval-runs/{run_id}` until its status reaches a terminal value.\n\nSend one of these flat request forms:\n\n- **Saved eval:** provide `eval_id`, `configuration_id`, and `agent_id`.\n- **Direct create-and-run:** provide `prompt`, `judges`, and `agent_id`. You may also provide a repository, write-only environment variables, and public skill packages.\n\nDo not combine saved-run and direct-run fields. No separate `mode` field is required."
+        "description": "Start an eval run\n\nQueues a durable eval job and returns immediately with a polling ID. Poll `GET /v2/eval-runs/{run_id}` until its status reaches a terminal value.\n\nSend one of these flat request forms:\n\n- **Saved eval:** provide `eval_id`; optionally replace its prompts, persona, agents, CLIs, MCPs, or runs for this launch. A saved configuration can still provide repository, environment, skills, and legacy tool selections.\n- **Direct create-and-run:** provide `prompts`, `judges`, and `agent`. You may also provide a persona, tags, repository, write-only environment variables, saved or inline CLIs and MCP servers, and skills.\n\nThe run fans out across prompts × agents × runs. Persona is shared metadata and does not multiply attempts. No separate `mode` field is required."
     },
     {
         "name": "evals-analysis",

package/esm/tool-names.js.map

@@ -1 +1 @@
-{"version":3,"file":"tool-names.js","sourceRoot":"","sources":["../src/tool-names.ts"],"names":[],"mappings":"AAAA,+BAA+B;AAC/B,MAAM,CAAC,MAAM,SAAS,GAAgD;IACpE;QACE,MAAM,EAAE,sBAAsB;QAC9B,aAAa,EAAE,6GAA6G;KAC7H;IACD;QACE,MAAM,EAAE,2BAA2B;QACnC,aAAa,EAAE,0HAA0H;KAC1I;IACD;QACE,MAAM,EAAE,2CAA2C;QACnD,aAAa,EAAE,oOAAoO;KACpP;IACD;QACE,MAAM,EAAE,4CAA4C;QACpD,aAAa,EAAE,oKAAoK;KACpL;IACD;QACE,MAAM,EAAE,yCAAyC;QACjD,aAAa,EAAE,iKAAiK;KACjL;IACD;QACE,MAAM,EAAE,2CAA2C;QACnD,aAAa,EAAE,wKAAwK;KACxL;IACD;QACE,MAAM,EAAE,yBAAyB;QACjC,aAAa,EAAE,qHAAqH;KACrI;IACD;QACE,MAAM,EAAE,2BAA2B;QACnC,aAAa,EAAE,6JAA6J;KAC7K;IACD;QACE,MAAM,EAAE,+BAA+B;QACvC,aAAa,EAAE,yEAAyE;KACzF;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,qEAAqE;KACrF;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,uEAAuE;KACvF;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,wJAAwJ;KACxK;IACD;QACE,MAAM,EAAE,6BAA6B;QACrC,aAAa,EAAE,wJAAwJ;KACxK;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,6FAA6F;KAC7G;IACD;QACE,MAAM,EAAE,6BAA6B;QACrC,aAAa,EAAE,4FAA4F;KAC5G;IACD;QACE,MAAM,EAAE,2BAA2B;QACnC,aAAa,EAAE,mHAAmH;KACnI;IACD;QACE,MAAM,EAAE,2BAA2B;QACnC,aAAa,EAAE,qHAAqH;KACrI;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,iMAAiM;KACjN;IACD;QACE,MAAM,EAAE,yCAAyC;QACjD,aAAa,EAAE,kHAAkH;KAClI;IACD;QACE,MAAM,EAAE,qBAAqB;QAC7B,aAAa,EAAE,0IAA0I;KAC1J;IACD;QACE,MAAM,EAAE,yBAAyB;QACjC,aAAa,EAAE,uEAAuE;KACvF;IACD;QACE,MAAM,EAAE,uBAAuB;QAC/B,aAAa,EAAE,4IAA4I;KAC5J;IACD;QACE,MAAM,EAAE,kCAAkC;QAC1C,aAAa,EAAE,6KAA6K;KAC7L;IACD;QACE,MAAM,EAAE,6BAA6B;QACrC,aAAa,EAAE,4HAA4H;KAC5I;IACD;QACE,MAAM,EAAE,6BAA6B;QACrC,aAAa,EAAE,uGAAuG;KACvH;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,6NAA6N;KAC7O;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,uHAAuH;KACvI;IACD;QACE,MAAM,EAAE,qBAAqB;QAC7B,aAAa,EAAE,iHAAiH;KACjI;IACD;QACE,MAAM,EAAE,uBAAuB;QAC/B,aAAa,EAAE,sJAAsJ;KACtK;IACD;QACE,MAAM,EAAE,yBAAyB;QACjC,aAAa,EAAE,gEAAgE;KAChF;IACD;QACE,MAAM,EAAE,uBAAuB;QAC/B,aAAa,EAAE,sHAAsH;KACtI;IACD;QACE,MAAM,EAAE,uBAAuB;QAC/B,aAAa,EAAE,iFAAiF;KACjG;IACD;QACE,MAAM,EAAE,mCAAmC;QAC3C,aAAa,EAAE,sIAAsI;KACtJ;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,sIAAsI;KACtJ;IACD;QACE,MAAM,EAAE,uCAAuC;QAC/C,aAAa,EAAE,wEAAwE;KACxF;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,yFAAyF;KACzG;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,qGAAqG;KACrH;IACD;QACE,MAAM,EAAE,WAAW;QACnB,aAAa,EAAE,giBAAgiB;KAChjB;IACD;QACE,MAAM,EAAE,gBAAgB;QACxB,aAAa,EAAE,wTAAwT;KACxU;IACD;QACE,MAAM,EAAE,aAAa;QACrB,aAAa,EAAE,+NAA+N;KAC/O;IACD;QACE,MAAM,EAAE,6BAA6B;QACrC,aAAa,EAAE,0HAA0H;KAC1I;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,iHAAiH;KACjI;IACD;QACE,MAAM,EAAE,mCAAmC;QAC3C,aAAa,EAAE,uJAAuJ;KACvK;IACD;QACE,MAAM,EAAE,uCAAuC;QAC/C,aAAa,EAAE,qGAAqG;KACrH;IACD;QACE,MAAM,EAAE,uBAAuB;QAC/B,aAAa,EAAE,8GAA8G;KAC9H;IACD;QACE,MAAM,EAAE,mCAAmC;QAC3C,aAAa,EAAE,6OAA6O;KAC7P;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,wFAAwF;KACxG;IACD;QACE,MAAM,EAAE,uCAAuC;QAC/C,aAAa,EAAE,mEAAmE;KACnF;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,uFAAuF;KACvG;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,mGAAmG;KACnH;IACD;QACE,MAAM,EAAE,mCAAmC;QAC3C,aAAa,EAAE,oFAAoF;KACpG;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,iGAAiG;KACjH;IACD;QACE,MAAM,EAAE,oBAAoB;QAC5B,aAAa,EAAE,qFAAqF;KACrG;IACD;QACE,MAAM,EAAE,uBAAuB;QAC/B,aAAa,EAAE,wFAAwF;KACxG;CACF,CAAC"}
+{"version":3,"file":"tool-names.js","sourceRoot":"","sources":["../src/tool-names.ts"],"names":[],"mappings":"AAAA,+BAA+B;AAC/B,MAAM,CAAC,MAAM,SAAS,GAAgD;IACpE;QACE,MAAM,EAAE,sBAAsB;QAC9B,aAAa,EAAE,6GAA6G;KAC7H;IACD;QACE,MAAM,EAAE,2BAA2B;QACnC,aAAa,EAAE,0HAA0H;KAC1I;IACD;QACE,MAAM,EAAE,2CAA2C;QACnD,aAAa,EAAE,oOAAoO;KACpP;IACD;QACE,MAAM,EAAE,4CAA4C;QACpD,aAAa,EAAE,oKAAoK;KACpL;IACD;QACE,MAAM,EAAE,yCAAyC;QACjD,aAAa,EAAE,iKAAiK;KACjL;IACD;QACE,MAAM,EAAE,2CAA2C;QACnD,aAAa,EAAE,wKAAwK;KACxL;IACD;QACE,MAAM,EAAE,yBAAyB;QACjC,aAAa,EAAE,qHAAqH;KACrI;IACD;QACE,MAAM,EAAE,2BAA2B;QACnC,aAAa,EAAE,6JAA6J;KAC7K;IACD;QACE,MAAM,EAAE,+BAA+B;QACvC,aAAa,EAAE,yEAAyE;KACzF;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,qEAAqE;KACrF;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,uEAAuE;KACvF;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,wJAAwJ;KACxK;IACD;QACE,MAAM,EAAE,6BAA6B;QACrC,aAAa,EAAE,wJAAwJ;KACxK;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,6FAA6F;KAC7G;IACD;QACE,MAAM,EAAE,6BAA6B;QACrC,aAAa,EAAE,4FAA4F;KAC5G;IACD;QACE,MAAM,EAAE,2BAA2B;QACnC,aAAa,EAAE,mHAAmH;KACnI;IACD;QACE,MAAM,EAAE,2BAA2B;QACnC,aAAa,EAAE,qHAAqH;KACrI;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,iMAAiM;KACjN;IACD;QACE,MAAM,EAAE,yCAAyC;QACjD,aAAa,EAAE,kHAAkH;KAClI;IACD;QACE,MAAM,EAAE,qBAAqB;QAC7B,aAAa,EAAE,0IAA0I;KAC1J;IACD;QACE,MAAM,EAAE,yBAAyB;QACjC,aAAa,EAAE,uEAAuE;KACvF;IACD;QACE,MAAM,EAAE,uBAAuB;QAC/B,aAAa,EAAE,4IAA4I;KAC5J;IACD;QACE,MAAM,EAAE,kCAAkC;QAC1C,aAAa,EAAE,6KAA6K;KAC7L;IACD;QACE,MAAM,EAAE,6BAA6B;QACrC,aAAa,EAAE,4HAA4H;KAC5I;IACD;QACE,MAAM,EAAE,6BAA6B;QACrC,aAAa,EAAE,uGAAuG;KACvH;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,6NAA6N;KAC7O;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,uHAAuH;KACvI;IACD;QACE,MAAM,EAAE,qBAAqB;QAC7B,aAAa,EAAE,8EAA8E;KAC9F;IACD;QACE,MAAM,EAAE,uBAAuB;QAC/B,aAAa,EAAE,mSAAmS;KACnT;IACD;QACE,MAAM,EAAE,yBAAyB;QACjC,aAAa,EAAE,mEAAmE;KACnF;IACD;QACE,MAAM,EAAE,uBAAuB;QAC/B,aAAa,EAAE,uNAAuN;KACvO;IACD;QACE,MAAM,EAAE,uBAAuB;QAC/B,aAAa,EAAE,iFAAiF;KACjG;IACD;QACE,MAAM,EAAE,mCAAmC;QAC3C,aAAa,EAAE,sIAAsI;KACtJ;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,sIAAsI;KACtJ;IACD;QACE,MAAM,EAAE,uCAAuC;QAC/C,aAAa,EAAE,wEAAwE;KACxF;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,yFAAyF;KACzG;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,qGAAqG;KACrH;IACD;QACE,MAAM,EAAE,WAAW;QACnB,aAAa,EAAE,uxBAAuxB;KACvyB;IACD;QACE,MAAM,EAAE,gBAAgB;QACxB,aAAa,EAAE,wTAAwT;KACxU;IACD;QACE,MAAM,EAAE,aAAa;QACrB,aAAa,EAAE,+NAA+N;KAC/O;IACD;QACE,MAAM,EAAE,6BAA6B;QACrC,aAAa,EAAE,0HAA0H;KAC1I;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,iHAAiH;KACjI;IACD;QACE,MAAM,EAAE,mCAAmC;QAC3C,aAAa,EAAE,uJAAuJ;KACvK;IACD;QACE,MAAM,EAAE,uCAAuC;QAC/C,aAAa,EAAE,qGAAqG;KACrH;IACD;QACE,MAAM,EAAE,uBAAuB;QAC/B,aAAa,EAAE,8GAA8G;KAC9H;IACD;QACE,MAAM,EAAE,mCAAmC;QAC3C,aAAa,EAAE,6OAA6O;KAC7P;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,wFAAwF;KACxG;IACD;QACE,MAAM,EAAE,uCAAuC;QAC/C,aAAa,EAAE,mEAAmE;KACnF;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,uFAAuF;KACvG;IACD;QACE,MAAM,EAAE,qCAAqC;QAC7C,aAAa,EAAE,mGAAmG;KACnH;IACD;QACE,MAAM,EAAE,mCAAmC;QAC3C,aAAa,EAAE,oFAAoF;KACpG;IACD;QACE,MAAM,EAAE,iCAAiC;QACzC,aAAa,EAAE,iGAAiG;KACjH;IACD;QACE,MAAM,EAAE,oBAAoB;QAC5B,aAAa,EAAE,qFAAqF;KACrG;IACD;QACE,MAAM,EAAE,uBAAuB;QAC/B,aAAa,EAAE,wFAAwF;KACxG;CACF,CAAC"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lightsage/mcp",
-  "version": "0.3.0",
+  "version": "0.3.1",
   "author": "Speakeasy",
   "type": "module",
   "sideEffects": false,

package/src/funcs/evalsRun.ts

@@ -34,10 +34,10 @@ import { Result } from "../types/fp.js";
  *
  * Send one of these flat request forms:
  *
- * - **Saved eval:** provide `eval_id`, `configuration_id`, and `agent_id`.
- * - **Direct create-and-run:** provide `prompt`, `judges`, and `agent_id`. You may also provide a repository, write-only environment variables, and public skill packages.
+ * - **Saved eval:** provide `eval_id`; optionally replace its prompts, persona, agents, CLIs, MCPs, or runs for this launch. A saved configuration can still provide repository, environment, skills, and legacy tool selections.
+ * - **Direct create-and-run:** provide `prompts`, `judges`, and `agent`. You may also provide a persona, tags, repository, write-only environment variables, saved or inline CLIs and MCP servers, and skills.
  *
- * Do not combine saved-run and direct-run fields. No separate `mode` field is required.
+ * The run fans out across prompts × agents × runs. Persona is shared metadata and does not multiply attempts. No separate `mode` field is required.
  */
 export function evalsRun(
   client$: LightsageMcpCore,

package/src/funcs/evalsV2EvalsCreate.ts

@@ -30,7 +30,7 @@ import { Result } from "../types/fp.js";
  * Create Eval
  *
  * @remarks
- * Creates an eval with a task prompt, subject, and plain-text success criterion in `judge`. The eval can then be run through eval runs.
+ * Creates a reusable simulation definition with one or more task prompts and judges. Optional agents, persona, CLI installations, MCP servers, repeat count, and tags are saved with the eval. Creating the eval does not start a run and does not accept secret environment values.
  */
 export function evalsV2EvalsCreate(
   client$: LightsageMcpCore,

package/src/funcs/evalsV2EvalsList.ts

@@ -24,7 +24,7 @@ import { Result } from "../types/fp.js";
  * List Evals
  *
  * @remarks
- * Returns all active evals including workflow, MCP, CLI, and skill evals owned by the organization.
+ * Returns all active eval definitions owned by the organization.
  */
 export function evalsV2EvalsList(
   client$: LightsageMcpCore,

package/src/funcs/evalsV2EvalsRetrieve.ts

@@ -30,7 +30,7 @@ import { Result } from "../types/fp.js";
  * Retrieve Eval
  *
  * @remarks
- * Returns one eval and its judge configuration.
+ * Returns one reusable eval simulation definition.
  */
 export function evalsV2EvalsRetrieve(
   client$: LightsageMcpCore,

package/src/funcs/evalsV2EvalsUpdate.ts

@@ -30,7 +30,7 @@ import { Result } from "../types/fp.js";
  * Update Eval
  *
  * @remarks
- * Updates an eval's task prompt, subject, plain-text `judge`, or name. Omitted fields remain unchanged.
+ * Updates a reusable eval simulation definition. Omitted fields remain unchanged. Empty `agent`, `clis`, `mcps`, or `tags` arrays clear those collections, and `persona: null` clears the saved persona.
  */
 export function evalsV2EvalsUpdate(
   client$: LightsageMcpCore,

package/src/landing-page.ts

@@ -931,7 +931,7 @@ http_headers = { "lightsage-api-key-auth" = "YOUR_LIGHTSAGE_API_KEY_AUTH" }`;
         <h1>Instructions</h1>
         <p>One-click installation for Claude Desktop users</p>
         <div class="instruction-item">
-          <a href="https://github.com/lightsagehq/mcp/releases/download/v0.3.0/mcp-server.mcpb" download="mcp-server.mcpb" class="action-button header-action" style="display: inline-flex; margin-bottom: 16px;">
+          <a href="https://github.com/lightsagehq/mcp/releases/download/v0.3.1/mcp-server.mcpb" download="mcp-server.mcpb" class="action-button header-action" style="display: inline-flex; margin-bottom: 16px;">
             📥 Download MCP Bundle
           </a>
         </div>

package/src/lib/config.ts

@@ -66,7 +66,7 @@ export function serverURLFromOptions(options: SDKOptions): URL | null {
 export const SDK_METADATA = {
   language: "typescript",
   openapiDocVersion: "1.0.0",
-  sdkVersion: "0.3.0",
-  genVersion: "2.935.1",
-  userAgent: "speakeasy-sdk/mcp-typescript 0.3.0 2.935.1 1.0.0 @lightsage/mcp",
+  sdkVersion: "0.3.1",
+  genVersion: "2.937.18",
+  userAgent: "speakeasy-sdk/mcp-typescript 0.3.1 2.937.18 1.0.0 @lightsage/mcp",
 } as const;

package/src/mcp-server/mcp-server.ts

@@ -22,7 +22,7 @@ const routes = buildRouteMap({
 export const app = buildApplication(routes, {
   name: "mcp",
   versionInfo: {
-    currentVersion: "0.3.0",
+    currentVersion: "0.3.1",
   },
 });
 

package/src/mcp-server/server.ts

@@ -92,7 +92,7 @@ export function createMCPServer(deps: {
       sizes: ["any"],
     }],
     name: "LightsageMcp",
-    version: "0.3.0",
+    version: "0.3.1",
   });
 
   const getClient = deps.getSDK || (() =>

package/src/mcp-server/tools/evalsRun.ts

@@ -19,10 +19,10 @@ Queues a durable eval job and returns immediately with a polling ID. Poll \`GET
 
 Send one of these flat request forms:
 
-- **Saved eval:** provide \`eval_id\`, \`configuration_id\`, and \`agent_id\`.
-- **Direct create-and-run:** provide \`prompt\`, \`judges\`, and \`agent_id\`. You may also provide a repository, write-only environment variables, and public skill packages.
+- **Saved eval:** provide \`eval_id\`; optionally replace its prompts, persona, agents, CLIs, MCPs, or runs for this launch. A saved configuration can still provide repository, environment, skills, and legacy tool selections.
+- **Direct create-and-run:** provide \`prompts\`, \`judges\`, and \`agent\`. You may also provide a persona, tags, repository, write-only environment variables, saved or inline CLIs and MCP servers, and skills.
 
-Do not combine saved-run and direct-run fields. No separate \`mode\` field is required.`,
+The run fans out across prompts × agents × runs. Persona is shared metadata and does not multiply attempts. No separate \`mode\` field is required.`,
   annotations: {
     "title": "",
     "destructiveHint": false,

package/src/mcp-server/tools/evalsV2EvalsCreate.ts

@@ -15,7 +15,7 @@ export const tool$evalsV2EvalsCreate: ToolDefinition<typeof args> = {
   name: "evals-v2-evals-create",
   description: `Create Eval
 
-Creates an eval with a task prompt, subject, and plain-text success criterion in \`judge\`. The eval can then be run through eval runs.`,
+Creates a reusable simulation definition with one or more task prompts and judges. Optional agents, persona, CLI installations, MCP servers, repeat count, and tags are saved with the eval. Creating the eval does not start a run and does not accept secret environment values.`,
   annotations: {
     "title": "",
     "destructiveHint": false,

package/src/mcp-server/tools/evalsV2EvalsList.ts

@@ -10,7 +10,7 @@ export const tool$evalsV2EvalsList: ToolDefinition = {
   name: "evals-v2-evals-list",
   description: `List Evals
 
-Returns all active evals including workflow, MCP, CLI, and skill evals owned by the organization.`,
+Returns all active eval definitions owned by the organization.`,
   annotations: {
     "title": "",
     "destructiveHint": false,

package/src/mcp-server/tools/evalsV2EvalsRetrieve.ts

@@ -15,7 +15,7 @@ export const tool$evalsV2EvalsRetrieve: ToolDefinition<typeof args> = {
   name: "evals-v2-evals-retrieve",
   description: `Retrieve Eval
 
-Returns one eval and its judge configuration.`,
+Returns one reusable eval simulation definition.`,
   annotations: {
     "title": "",
     "destructiveHint": false,

package/src/mcp-server/tools/evalsV2EvalsUpdate.ts

@@ -15,7 +15,7 @@ export const tool$evalsV2EvalsUpdate: ToolDefinition<typeof args> = {
   name: "evals-v2-evals-update",
   description: `Update Eval
 
-Updates an eval's task prompt, subject, plain-text \`judge\`, or name. Omitted fields remain unchanged.`,
+Updates a reusable eval simulation definition. Omitted fields remain unchanged. Empty \`agent\`, \`clis\`, \`mcps\`, or \`tags\` arrays clear those collections, and \`persona: null\` clears the saved persona.`,
   annotations: {
     "title": "",
     "destructiveHint": false,

package/src/models/createv2evalrequest.ts

@@ -4,42 +4,86 @@
  */
 
 import * as z from "zod";
-import { ClosedEnum } from "../types/enums.js";
+import {
+  V2InlineCliReference,
+  V2InlineCliReference$zodSchema,
+} from "./v2inlineclireference.js";
+import {
+  V2InlineMcpDefinition,
+  V2InlineMcpDefinition$zodSchema,
+} from "./v2inlinemcpdefinition.js";
+import {
+  V2SavedCliReference,
+  V2SavedCliReference$zodSchema,
+} from "./v2savedclireference.js";
+import {
+  V2SavedMcpReference,
+  V2SavedMcpReference$zodSchema,
+} from "./v2savedmcpreference.js";
 
-export const CreateV2EvalRequestType = {
-  Workflow: "workflow",
-  Mcp: "mcp",
-  Cli: "cli",
-  Skill: "skill",
-} as const;
-export type CreateV2EvalRequestType = ClosedEnum<
-  typeof CreateV2EvalRequestType
->;
+export type CreateV2EvalRequestCli = V2SavedCliReference | V2InlineCliReference;
 
-export const CreateV2EvalRequestType$zodSchema = z.enum([
-  "workflow",
-  "mcp",
-  "cli",
-  "skill",
+export const CreateV2EvalRequestCli$zodSchema: z.ZodType<
+  CreateV2EvalRequestCli
+> = z.union([
+  V2SavedCliReference$zodSchema,
+  V2InlineCliReference$zodSchema,
+]);
+
+export type CreateV2EvalRequestMcp =
+  | V2InlineMcpDefinition
+  | V2SavedMcpReference;
+
+export const CreateV2EvalRequestMcp$zodSchema: z.ZodType<
+  CreateV2EvalRequestMcp
+> = z.union([
+  V2InlineMcpDefinition$zodSchema,
+  V2SavedMcpReference$zodSchema,
 ]);
 
 export type CreateV2EvalRequest = {
   name: string;
-  type: CreateV2EvalRequestType;
-  prompt: string;
-  judge: string;
-  mcp_server_id?: string | undefined;
-  cli_id?: string | undefined;
-  skill_id?: string | undefined;
+  prompts: Array<string>;
+  judges: Array<string>;
+  agent?: Array<string> | undefined;
+  persona?: string | undefined;
+  clis?: Array<V2SavedCliReference | V2InlineCliReference> | undefined;
+  mcps?: Array<V2InlineMcpDefinition | V2SavedMcpReference> | undefined;
+  runs?: number | undefined;
+  tags?: Array<string> | undefined;
 };
 
 export const CreateV2EvalRequest$zodSchema: z.ZodType<CreateV2EvalRequest> = z
   .object({
-    cli_id: z.string().optional(),
-    judge: z.string().describe("Plain-text outcome that counts as success."),
-    mcp_server_id: z.string().optional(),
+    agent: z.array(z.string()).optional().describe(
+      "Runnable agents from `GET /v2/agents`. An empty array saves the eval without default agents. CLI clients expose repeatable `--agent`.",
+    ),
+    clis: z.array(z.union([
+      V2SavedCliReference$zodSchema,
+      V2InlineCliReference$zodSchema,
+    ])).optional().describe(
+      "Saved CLI references or inline installation commands.",
+    ),
+    judges: z.array(z.string()).describe(
+      "Plain-text success criteria. CLI clients expose this as repeatable `--judge`.",
+    ),
+    mcps: z.array(z.union([
+      V2InlineMcpDefinition$zodSchema,
+      V2SavedMcpReference$zodSchema,
+    ])).optional().describe(
+      "Saved MCP references or inline portable MCP definitions.",
+    ),
     name: z.string(),
-    prompt: z.string(),
-    skill_id: z.string().optional(),
-    type: CreateV2EvalRequestType$zodSchema,
+    persona: z.string().optional().describe(
+      "Optional perspective applied to the whole eval. It is stored as metadata and does not change prompts or multiply attempts.",
+    ),
+    prompts: z.array(z.string()).describe(
+      "Task prompts. CLI clients expose this as repeatable `--prompt`.",
+    ),
+    runs: z.int().default(1).describe(
+      "Independent attempts for every prompt-agent pair.",
+    ),
+    tags: z.array(z.string()).optional().describe(
+      "Optional labels for organizing the eval. CLI clients may expose this string array as a repeatable `--tag` flag.",
+    ),
   });

package/src/models/createv2evalrunrequest.ts

@@ -4,48 +4,120 @@
  */
 
 import * as z from "zod";
+import {
+  V2InlineCliReference,
+  V2InlineCliReference$zodSchema,
+} from "./v2inlineclireference.js";
+import {
+  V2InlineMcpDefinition,
+  V2InlineMcpDefinition$zodSchema,
+} from "./v2inlinemcpdefinition.js";
+import {
+  V2SavedCliReference,
+  V2SavedCliReference$zodSchema,
+} from "./v2savedclireference.js";
+import {
+  V2SavedMcpReference,
+  V2SavedMcpReference$zodSchema,
+} from "./v2savedmcpreference.js";
+
+export type CreateV2EvalRunRequestCli =
+  | V2SavedCliReference
+  | V2InlineCliReference;
+
+export const CreateV2EvalRunRequestCli$zodSchema: z.ZodType<
+  CreateV2EvalRunRequestCli
+> = z.union([
+  V2SavedCliReference$zodSchema,
+  V2InlineCliReference$zodSchema,
+]);
+
+export type CreateV2EvalRunRequestMcp =
+  | V2InlineMcpDefinition
+  | V2SavedMcpReference;
+
+export const CreateV2EvalRunRequestMcp$zodSchema: z.ZodType<
+  CreateV2EvalRunRequestMcp
+> = z.union([
+  V2InlineMcpDefinition$zodSchema,
+  V2SavedMcpReference$zodSchema,
+]);
 
 /**
- * Start an eval run using one of two flat request forms. Saved runs require `eval_id`, `configuration_id`, and `agent_id`. Direct create-and-run requests require `prompt`, at least one item in `judges`, and `agent_id`. Do not combine saved-run and direct-run fields; no separate mode field is required.
+ * Start an eval run using one of two flat request forms. Saved runs require `eval_id` and use saved simulation defaults unless the request replaces them. Direct create-and-run requests require `prompts`, `judges`, and `agent`. No separate mode field is required.
  */
 export type CreateV2EvalRunRequest = {
   eval_id?: string | undefined;
   configuration_id?: string | undefined;
-  agent_id: string;
-  prompt?: string | undefined;
-  judges?: Array<string> | undefined;
+  name?: string | undefined;
+  prompts?: Array<string> | null | undefined;
+  judges?: Array<string> | null | undefined;
+  agent?: Array<string> | null | undefined;
+  persona?: string | undefined;
+  clis?: Array<V2SavedCliReference | V2InlineCliReference> | null | undefined;
+  mcps?: Array<V2InlineMcpDefinition | V2SavedMcpReference> | null | undefined;
+  runs?: number | undefined;
+  tags?: Array<string> | undefined;
   repository?: string | undefined;
   env?: { [k: string]: string } | undefined;
   skills?: Array<string> | undefined;
+  skill_ids?: Array<string> | undefined;
 };
 
 export const CreateV2EvalRunRequest$zodSchema: z.ZodType<
   CreateV2EvalRunRequest
 > = z.object({
-  agent_id: z.string().describe(
-    "Required in both request forms. Runnable agent ID from `GET /v2/agents`.",
+  agent: z.array(z.string()).nullable().optional().describe(
+    "Runnable agents from `GET /v2/agents`. Required unless the saved eval already has agents. CLI clients expose repeatable `--agent`.",
+  ),
+  clis: z.array(z.union([
+    V2SavedCliReference$zodSchema,
+    V2InlineCliReference$zodSchema,
+  ])).nullable().optional().describe(
+    "Saved CLI references or inline installation commands. In saved mode, an explicit empty array disables saved CLIs for this launch.",
   ),
   configuration_id: z.string().optional().describe(
-    "Saved-run form only. Saved runtime configuration; required when `eval_id` is provided.",
+    "Saved-run form only. Optional saved repository, environment, skills, and legacy tool configuration.",
   ),
   env: z.record(z.string(), z.string()).optional().describe(
-    "Direct create-and-run form only. Write-only environment variables; values are stored only in the queued job's private configuration.",
+    "Write-only environment variables for this launch. Values are stored only in the queued job's private configuration.",
   ),
   eval_id: z.string().optional().describe(
     "Saved-run form only. Existing eval definition to run. Omit this field to use the direct create-and-run form.",
   ),
-  judges: z.array(z.string()).optional().describe(
-    "Direct create-and-run form only. Provide at least one success criterion. CLI clients may expose this string array as a repeatable `--judge` flag.",
+  judges: z.array(z.string()).nullable().optional().describe(
+    "Success criteria. Required in direct mode; in saved mode, replaces saved judges for this launch. CLI clients expose repeatable `--judge`.",
+  ),
+  mcps: z.array(z.union([
+    V2InlineMcpDefinition$zodSchema,
+    V2SavedMcpReference$zodSchema,
+  ])).nullable().optional().describe(
+    "Saved MCP references or inline portable definitions. In saved mode, an explicit empty array disables saved MCPs for this launch.",
   ),
-  prompt: z.string().optional().describe(
-    "Direct create-and-run form only. Task prompt; required when `eval_id` is omitted.",
+  name: z.string().optional().describe(
+    "Direct create-and-run form only. Display name for the durable eval. Defaults to the first prompt.",
+  ),
+  persona: z.string().optional().describe(
+    "Optional perspective for every attempt in this launch. It does not modify prompts or add a fan-out dimension.",
+  ),
+  prompts: z.array(z.string()).nullable().optional().describe(
+    "Task prompts. Required in direct mode; in saved mode, replaces saved prompts for this launch. CLI clients expose repeatable `--prompt`.",
   ),
   repository: z.string().optional().describe(
     "Direct create-and-run form only. A saved repository ID, built-in alias such as `nextjs-starter`, or a public GitHub HTTPS URL.",
   ),
+  runs: z.int().optional().describe(
+    "Independent attempts for every prompt-agent pair. Defaults to the saved eval value or 1 for a direct run.",
+  ),
+  skill_ids: z.array(z.string()).optional().describe(
+    "Direct create-and-run form only. Reusable skill IDs from `GET /v2/skills` to install before the agent starts.",
+  ),
   skills: z.array(z.string()).optional().describe(
     "Additional public skill packages to install before the agent starts in either request form, using GitHub `owner/repository` references such as `resend/resend-skills`. CLI clients may expose this string array as a repeatable `--skill` flag.",
   ),
+  tags: z.array(z.string()).optional().describe(
+    "Direct create-and-run form only. Optional labels saved on the created eval. CLI clients may expose this string array as a repeatable `--tag` flag.",
+  ),
 }).describe(
-  "Start an eval run using one of two flat request forms. Saved runs require `eval_id`, `configuration_id`, and `agent_id`. Direct create-and-run requests require `prompt`, at least one item in `judges`, and `agent_id`. Do not combine saved-run and direct-run fields; no separate mode field is required.",
+  "Start an eval run using one of two flat request forms. Saved runs require `eval_id` and use saved simulation defaults unless the request replaces them. Direct create-and-run requests require `prompts`, `judges`, and `agent`. No separate mode field is required.",
 );

package/src/models/updatev2evalrequest.ts

@@ -4,44 +4,76 @@
  */
 
 import * as z from "zod";
-import { ClosedEnum } from "../types/enums.js";
+import {
+  V2InlineCliReference,
+  V2InlineCliReference$zodSchema,
+} from "./v2inlineclireference.js";
+import {
+  V2InlineMcpDefinition,
+  V2InlineMcpDefinition$zodSchema,
+} from "./v2inlinemcpdefinition.js";
+import {
+  V2SavedCliReference,
+  V2SavedCliReference$zodSchema,
+} from "./v2savedclireference.js";
+import {
+  V2SavedMcpReference,
+  V2SavedMcpReference$zodSchema,
+} from "./v2savedmcpreference.js";
 
-export const UpdateV2EvalRequestType = {
-  Workflow: "workflow",
-  Mcp: "mcp",
-  Cli: "cli",
-  Skill: "skill",
-} as const;
-export type UpdateV2EvalRequestType = ClosedEnum<
-  typeof UpdateV2EvalRequestType
->;
+export type UpdateV2EvalRequestCli = V2SavedCliReference | V2InlineCliReference;
 
-export const UpdateV2EvalRequestType$zodSchema = z.enum([
-  "workflow",
-  "mcp",
-  "cli",
-  "skill",
+export const UpdateV2EvalRequestCli$zodSchema: z.ZodType<
+  UpdateV2EvalRequestCli
+> = z.union([
+  V2SavedCliReference$zodSchema,
+  V2InlineCliReference$zodSchema,
+]);
+
+export type UpdateV2EvalRequestMcp =
+  | V2InlineMcpDefinition
+  | V2SavedMcpReference;
+
+export const UpdateV2EvalRequestMcp$zodSchema: z.ZodType<
+  UpdateV2EvalRequestMcp
+> = z.union([
+  V2InlineMcpDefinition$zodSchema,
+  V2SavedMcpReference$zodSchema,
 ]);
 
 export type UpdateV2EvalRequest = {
   name?: string | undefined;
-  type?: UpdateV2EvalRequestType | undefined;
-  prompt?: string | undefined;
-  judge?: string | undefined;
-  mcp_server_id?: string | undefined;
-  cli_id?: string | undefined;
-  skill_id?: string | undefined;
+  prompts?: Array<string> | null | undefined;
+  judges?: Array<string> | null | undefined;
+  agent?: Array<string> | null | undefined;
+  persona?: string | undefined;
+  clis?: Array<V2SavedCliReference | V2InlineCliReference> | null | undefined;
+  mcps?: Array<V2InlineMcpDefinition | V2SavedMcpReference> | null | undefined;
+  runs?: number | undefined;
+  tags?: Array<string> | null | undefined;
 };
 
 export const UpdateV2EvalRequest$zodSchema: z.ZodType<UpdateV2EvalRequest> = z
   .object({
-    cli_id: z.string().optional(),
-    judge: z.string().optional().describe(
-      "Plain-text outcome that counts as success.",
+    agent: z.array(z.string()).nullable().optional(),
+    clis: z.array(z.union([
+      V2SavedCliReference$zodSchema,
+      V2InlineCliReference$zodSchema,
+    ])).nullable().optional(),
+    judges: z.array(z.string()).nullable().optional().describe(
+      "Replacement success criteria for the eval.",
     ),
-    mcp_server_id: z.string().optional(),
+    mcps: z.array(z.union([
+      V2InlineMcpDefinition$zodSchema,
+      V2SavedMcpReference$zodSchema,
+    ])).nullable().optional(),
     name: z.string().optional(),
-    prompt: z.string().optional(),
-    skill_id: z.string().optional(),
-    type: UpdateV2EvalRequestType$zodSchema.optional(),
+    persona: z.string().optional().describe(
+      "Replacement eval persona. Pass null to clear it; omit it to leave the saved value unchanged.",
+    ),
+    prompts: z.array(z.string()).nullable().optional(),
+    runs: z.int().optional(),
+    tags: z.array(z.string()).nullable().optional().describe(
+      "Replacement labels for the eval. Pass an empty array to clear all tags.",
+    ),
   });