@lightsage/mcp 0.3.0 → 0.3.1

package/README.md

@@ -33,9 +33,9 @@ Lightsage Public API: Public API for Lightsage CLI and external integrations.
 <details>
 <summary>Claude Desktop</summary>
 
-Install the MCP server as a Desktop Extension using the pre-built [`mcp-server.mcpb`](https://github.com/lightsagehq/mcp/releases/download/v0.3.0/mcp-server.mcpb) file:
+Install the MCP server as a Desktop Extension using the pre-built [`mcp-server.mcpb`](https://github.com/lightsagehq/mcp/releases/download/v0.3.1/mcp-server.mcpb) file:
 
-Simply drag and drop the [`mcp-server.mcpb`](https://github.com/lightsagehq/mcp/releases/download/v0.3.0/mcp-server.mcpb) file onto Claude Desktop to install the extension.
+Simply drag and drop the [`mcp-server.mcpb`](https://github.com/lightsagehq/mcp/releases/download/v0.3.1/mcp-server.mcpb) file onto Claude Desktop to install the extension.
 
 The MCP bundle package includes the MCP server and all necessary configuration. Once installed, the server will be available without additional setup.
 

package/bin/mcp-server.js

@@ -1453,7 +1453,7 @@ var init_versions = __esm(() => {
   version2 = {
     major: 4,
     minor: 6,
-    patch: 4
+    patch: 5
   };
 });
 
@@ -52769,9 +52769,9 @@ var init_config = __esm(() => {
   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"
   };
 });
 
@@ -54958,13 +54958,16 @@ var V2EvalResultSummary$zodSchema;
 var init_v2evalresultsummary = __esm(() => {
   init_zod();
   V2EvalResultSummary$zodSchema = object({
-    agent_id: string2().nullable().optional().describe("Runnable agent ID combining the harness and model."),
+    agent: string2().nullable().optional().describe("Runnable agent combining the harness and model."),
     created_at: string2().nullable().optional(),
     eval_id: string2().nullable().optional(),
     failure_reasons: array(string2()).nullable().optional(),
     id: string2().describe("Result ID, not an eval job ID."),
     latency_ms: int().nullable().optional(),
     passed: boolean2().nullable().optional(),
+    prompt: string2().nullable().optional(),
+    prompt_index: int().default(0),
+    repeat_index: int().default(0),
     score: number2().nullable().optional().describe("Score from 0 to 1.")
   });
 });
@@ -55114,7 +55117,7 @@ var V2EvalResult$zodSchema;
 var init_v2evalresult = __esm(() => {
   init_zod();
   V2EvalResult$zodSchema = object({
-    agent_id: string2().nullable().optional().describe("Runnable agent ID combining the harness and model."),
+    agent: string2().nullable().optional().describe("Runnable agent combining the harness and model."),
     created_at: string2().nullable().optional(),
     eval_id: string2().nullable().optional(),
     execution_stderr: string2().nullable().optional(),
@@ -55126,6 +55129,8 @@ var init_v2evalresult = __esm(() => {
     output: string2().nullable().optional(),
     passed: boolean2().nullable().optional(),
     prompt: string2().nullable().optional(),
+    prompt_index: int().default(0),
+    repeat_index: int().default(0),
     score: number2().nullable().optional().describe("Score from 0 to 1."),
     tool_calls: array(record(string2(), any())).nullable().optional()
   });
@@ -55281,8 +55286,8 @@ var init_v2evalrun = __esm(() => {
   init_zod();
   init_publicevalrunprogress();
   V2EvalRun$zodSchema = object({
-    agent_id: string2().nullable().optional().describe("Runnable agent ID selected for this run."),
-    agent_name: string2().nullable().optional().describe("Display name captured for the selected agent."),
+    agent: array(string2()).optional().describe("Runnable agents selected for this simulation."),
+    attempt_count: int().default(1).describe("Total concrete attempts after all fan-out dimensions."),
     completed_at: string2().nullable().optional(),
     configuration_id: string2().nullable().optional(),
     created_at: string2().nullable().optional(),
@@ -55293,7 +55298,10 @@ var init_v2evalrun = __esm(() => {
     eval_name: string2().nullable().optional().describe("Display name captured for the eval when the run was created."),
     id: string2().describe("Run ID to use when polling eval-run status."),
     job_id: string2().describe("Alias of `id` for the durable parent job."),
+    persona: string2().nullable().optional().describe("Perspective applied to every attempt in this simulation."),
     progress: PublicEvalRunProgress$zodSchema.optional(),
+    prompt_count: int().default(1).describe("Number of prompts included in the simulation."),
+    runs: int().default(1).describe("Independent attempts requested for every prompt-agent pair."),
     stage: string2().nullable().optional(),
     started_at: string2().nullable().optional(),
     status: string2()
@@ -55574,13 +55582,15 @@ var V2EvalRunAnalysis$zodSchema;
 var init_v2evalrunanalysis = __esm(() => {
   init_zod();
   V2EvalRunAnalysis$zodSchema = object({
-    agent_id: string2().nullable().optional().describe("Runnable agent ID used for this concrete attempt."),
+    agent: string2().nullable().optional().describe("Runnable agent used for this concrete attempt."),
     analysis: record(string2(), any()).optional().describe("Persisted eval_run_analysis payload, including verdict, outcome, findings, bottleneck, and analysis sections."),
     analysis_version: string2().nullable().optional().describe("Version of the analysis pipeline that produced this payload."),
     attempt_id: string2().describe("Child eval-attempt ID used to correlate the analysis with trace events."),
     created_at: string2().nullable().optional(),
     eval_id: string2().nullable().optional().describe("Public eval definition ID, not the backing executable eval ID."),
     id: string2().describe("Persisted eval analysis ID."),
+    prompt: string2().nullable().optional().describe("Exact task prompt used for this concrete attempt."),
+    prompt_index: int().default(0).describe("Zero-based prompt index within the submitted simulation."),
     repeat_index: int().default(0).describe("Zero-based repeat index within the submitted run."),
     result_id: string2().describe("Concrete eval result ID analyzed for this attempt."),
     updated_at: string2().nullable().optional(),
@@ -55735,20 +55745,92 @@ Returns the persisted analysis produced for each completed attempt in one eval r
   };
 });
 
+// src/models/v2inlineclireference.ts
+var V2InlineCliReference$zodSchema;
+var init_v2inlineclireference = __esm(() => {
+  init_zod();
+  V2InlineCliReference$zodSchema = object({
+    install_command: string2().describe("Shell command to install this CLI before the agent starts.")
+  });
+});
+
+// src/models/v2inlinemcpdefinition.ts
+var Type$zodSchema, V2InlineMcpDefinition$zodSchema;
+var init_v2inlinemcpdefinition = __esm(() => {
+  init_zod();
+  Type$zodSchema = _enum([
+    "http",
+    "sse",
+    "stdio"
+  ]);
+  V2InlineMcpDefinition$zodSchema = object({
+    args: array(string2()).optional(),
+    command: string2().nullable().optional().describe("Executable command for a `stdio` MCP server."),
+    env: record(string2(), string2()).optional().describe("Environment mapping for a `stdio` MCP server. Values may reference run secrets with `${VARIABLE_NAME}`."),
+    headers: record(string2(), string2()).optional().describe("HTTP headers. Reference run secrets with `${VARIABLE_NAME}` instead of embedding credentials."),
+    name: string2().describe("Stable server name used in the selected agent's MCP config."),
+    type: Type$zodSchema,
+    url: string2().nullable().optional().describe("HTTPS endpoint for an `http` or `sse` MCP server.")
+  });
+});
+
+// src/models/v2savedclireference.ts
+var V2SavedCliReference$zodSchema;
+var init_v2savedclireference = __esm(() => {
+  init_zod();
+  V2SavedCliReference$zodSchema = object({
+    id: string2().describe("Saved CLI ID from `GET /v2/clis`.")
+  });
+});
+
+// src/models/v2savedmcpreference.ts
+var V2SavedMcpReference$zodSchema;
+var init_v2savedmcpreference = __esm(() => {
+  init_zod();
+  V2SavedMcpReference$zodSchema = object({
+    id: string2().describe("Saved MCP server ID from `GET /v2/mcp-servers`.")
+  });
+});
+
 // src/models/createv2evalrunrequest.ts
-var CreateV2EvalRunRequest$zodSchema;
+var CreateV2EvalRunRequestCli$zodSchema, CreateV2EvalRunRequestMcp$zodSchema, CreateV2EvalRunRequest$zodSchema;
 var init_createv2evalrunrequest = __esm(() => {
   init_zod();
+  init_v2inlineclireference();
+  init_v2inlinemcpdefinition();
+  init_v2savedclireference();
+  init_v2savedmcpreference();
+  CreateV2EvalRunRequestCli$zodSchema = union([
+    V2SavedCliReference$zodSchema,
+    V2InlineCliReference$zodSchema
+  ]);
+  CreateV2EvalRunRequestMcp$zodSchema = union([
+    V2InlineMcpDefinition$zodSchema,
+    V2SavedMcpReference$zodSchema
+  ]);
   CreateV2EvalRunRequest$zodSchema = object({
-    agent_id: string2().describe("Required in both request forms. Runnable agent ID from `GET /v2/agents`."),
-    configuration_id: string2().optional().describe("Saved-run form only. Saved runtime configuration; required when `eval_id` is provided."),
-    env: record(string2(), string2()).optional().describe("Direct create-and-run form only. Write-only environment variables; values are stored only in the queued job's private configuration."),
+    agent: array(string2()).nullable().optional().describe("Runnable agents from `GET /v2/agents`. Required unless the saved eval already has agents. CLI clients expose repeatable `--agent`."),
+    clis: array(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: string2().optional().describe("Saved-run form only. Optional saved repository, environment, skills, and legacy tool configuration."),
+    env: record(string2(), string2()).optional().describe("Write-only environment variables for this launch. Values are stored only in the queued job's private configuration."),
     eval_id: string2().optional().describe("Saved-run form only. Existing eval definition to run. Omit this field to use the direct create-and-run form."),
-    judges: array(string2()).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."),
-    prompt: string2().optional().describe("Direct create-and-run form only. Task prompt; required when `eval_id` is omitted."),
+    judges: array(string2()).nullable().optional().describe("Success criteria. Required in direct mode; in saved mode, replaces saved judges for this launch. CLI clients expose repeatable `--judge`."),
+    mcps: array(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."),
+    name: string2().optional().describe("Direct create-and-run form only. Display name for the durable eval. Defaults to the first prompt."),
+    persona: string2().optional().describe("Optional perspective for every attempt in this launch. It does not modify prompts or add a fan-out dimension."),
+    prompts: array(string2()).nullable().optional().describe("Task prompts. Required in direct mode; in saved mode, replaces saved prompts for this launch. CLI clients expose repeatable `--prompt`."),
     repository: string2().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."),
-    skills: array(string2()).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.")
-  }).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.");
+    runs: int().optional().describe("Independent attempts for every prompt-agent pair. Defaults to the saved eval value or 1 for a direct run."),
+    skill_ids: array(string2()).optional().describe("Direct create-and-run form only. Reusable skill IDs from `GET /v2/skills` to install before the agent starts."),
+    skills: array(string2()).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: array(string2()).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` 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.");
 });
 
 // src/funcs/evalsRun.ts
@@ -55841,10 +55923,10 @@ 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,
@@ -55871,8 +55953,7 @@ var V2EvalRunTraceEvent$zodSchema;
 var init_v2evalruntraceevent = __esm(() => {
   init_zod();
   V2EvalRunTraceEvent$zodSchema = object({
-    agent_id: string2().nullable().optional(),
-    agent_name: string2().nullable().optional(),
+    agent: string2().nullable().optional(),
     category: string2(),
     child_job_id: string2().describe("Child eval-attempt ID associated with the event."),
     configuration_id: string2().nullable().optional(),
@@ -55881,6 +55962,8 @@ var init_v2evalruntraceevent = __esm(() => {
     eval_name: string2().nullable().optional(),
     event_id: int(),
     message: string2(),
+    prompt_index: int().default(0),
+    repeat_index: int().default(0),
     run_id: string2(),
     timestamp: string2().nullable().optional(),
     type: string2()
@@ -56759,23 +56842,37 @@ Updates an eval configuration; omitted fields remain unchanged.`,
 });
 
 // src/models/createv2evalrequest.ts
-var CreateV2EvalRequestType$zodSchema, CreateV2EvalRequest$zodSchema;
+var CreateV2EvalRequestCli$zodSchema, CreateV2EvalRequestMcp$zodSchema, CreateV2EvalRequest$zodSchema;
 var init_createv2evalrequest = __esm(() => {
   init_zod();
-  CreateV2EvalRequestType$zodSchema = _enum([
-    "workflow",
-    "mcp",
-    "cli",
-    "skill"
+  init_v2inlineclireference();
+  init_v2inlinemcpdefinition();
+  init_v2savedclireference();
+  init_v2savedmcpreference();
+  CreateV2EvalRequestCli$zodSchema = union([
+    V2SavedCliReference$zodSchema,
+    V2InlineCliReference$zodSchema
+  ]);
+  CreateV2EvalRequestMcp$zodSchema = union([
+    V2InlineMcpDefinition$zodSchema,
+    V2SavedMcpReference$zodSchema
   ]);
   CreateV2EvalRequest$zodSchema = object({
-    cli_id: string2().optional(),
-    judge: string2().describe("Plain-text outcome that counts as success."),
-    mcp_server_id: string2().optional(),
+    agent: array(string2()).optional().describe("Runnable agents from `GET /v2/agents`. An empty array saves the eval without default agents. CLI clients expose repeatable `--agent`."),
+    clis: array(union([
+      V2SavedCliReference$zodSchema,
+      V2InlineCliReference$zodSchema
+    ])).optional().describe("Saved CLI references or inline installation commands."),
+    judges: array(string2()).describe("Plain-text success criteria. CLI clients expose this as repeatable `--judge`."),
+    mcps: array(union([
+      V2InlineMcpDefinition$zodSchema,
+      V2SavedMcpReference$zodSchema
+    ])).optional().describe("Saved MCP references or inline portable MCP definitions."),
     name: string2(),
-    prompt: string2(),
-    skill_id: string2().optional(),
-    type: CreateV2EvalRequestType$zodSchema
+    persona: string2().optional().describe("Optional perspective applied to the whole eval. It is stored as metadata and does not change prompts or multiply attempts."),
+    prompts: array(string2()).describe("Task prompts. CLI clients expose this as repeatable `--prompt`."),
+    runs: int().default(1).describe("Independent attempts for every prompt-agent pair."),
+    tags: array(string2()).optional().describe("Optional labels for organizing the eval. CLI clients may expose this string array as a repeatable `--tag` flag.")
   });
 });
 
@@ -56865,7 +56962,7 @@ var init_evalsV2EvalsCreate2 = __esm(() => {
     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,
@@ -56887,36 +56984,40 @@ Creates an eval with a task prompt, subject, and plain-text success criterion in
   };
 });
 
-// src/models/v2evaljudge.ts
-var V2EvalJudge$zodSchema;
-var init_v2evaljudge = __esm(() => {
-  init_zod();
-  V2EvalJudge$zodSchema = object({
-    criteria: string2().describe("The outcome that counts as success.")
-  });
-});
-
 // src/models/v2eval.ts
-var V2EvalType$zodSchema, V2Eval$zodSchema;
+var V2EvalCli$zodSchema, V2EvalMcp$zodSchema, V2Eval$zodSchema;
 var init_v2eval = __esm(() => {
   init_zod();
-  init_v2evaljudge();
-  V2EvalType$zodSchema = _enum([
-    "workflow",
-    "mcp",
-    "cli",
-    "skill"
+  init_v2inlineclireference();
+  init_v2inlinemcpdefinition();
+  init_v2savedclireference();
+  init_v2savedmcpreference();
+  V2EvalCli$zodSchema = union([
+    V2SavedCliReference$zodSchema,
+    V2InlineCliReference$zodSchema
+  ]);
+  V2EvalMcp$zodSchema = union([
+    V2InlineMcpDefinition$zodSchema,
+    V2SavedMcpReference$zodSchema
   ]);
   V2Eval$zodSchema = object({
-    cli_id: string2().nullable().optional(),
+    agent: array(string2()).optional(),
+    clis: array(union([
+      V2SavedCliReference$zodSchema,
+      V2InlineCliReference$zodSchema
+    ])).optional(),
     created_at: string2().nullable().optional(),
     id: string2(),
-    judge: V2EvalJudge$zodSchema,
-    mcp_server_id: string2().nullable().optional(),
+    judges: array(string2()).optional(),
+    mcps: array(union([
+      V2InlineMcpDefinition$zodSchema,
+      V2SavedMcpReference$zodSchema
+    ])).optional(),
     name: string2(),
-    prompt: string2(),
-    skill_id: string2().nullable().optional(),
-    type: V2EvalType$zodSchema,
+    persona: string2().nullable().optional(),
+    prompts: array(string2()).optional(),
+    runs: int().default(1),
+    tags: array(string2()).optional(),
     updated_at: string2().nullable().optional()
   });
 });
@@ -57120,7 +57221,7 @@ var init_evalsV2EvalsList2 = __esm(() => {
     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,
@@ -57247,7 +57348,7 @@ var init_evalsV2EvalsRetrieve2 = __esm(() => {
     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,
@@ -57270,23 +57371,37 @@ Returns one eval and its judge configuration.`,
 });
 
 // src/models/updatev2evalrequest.ts
-var UpdateV2EvalRequestType$zodSchema, UpdateV2EvalRequest$zodSchema;
+var UpdateV2EvalRequestCli$zodSchema, UpdateV2EvalRequestMcp$zodSchema, UpdateV2EvalRequest$zodSchema;
 var init_updatev2evalrequest = __esm(() => {
   init_zod();
-  UpdateV2EvalRequestType$zodSchema = _enum([
-    "workflow",
-    "mcp",
-    "cli",
-    "skill"
+  init_v2inlineclireference();
+  init_v2inlinemcpdefinition();
+  init_v2savedclireference();
+  init_v2savedmcpreference();
+  UpdateV2EvalRequestCli$zodSchema = union([
+    V2SavedCliReference$zodSchema,
+    V2InlineCliReference$zodSchema
+  ]);
+  UpdateV2EvalRequestMcp$zodSchema = union([
+    V2InlineMcpDefinition$zodSchema,
+    V2SavedMcpReference$zodSchema
   ]);
   UpdateV2EvalRequest$zodSchema = object({
-    cli_id: string2().optional(),
-    judge: string2().optional().describe("Plain-text outcome that counts as success."),
-    mcp_server_id: string2().optional(),
+    agent: array(string2()).nullable().optional(),
+    clis: array(union([
+      V2SavedCliReference$zodSchema,
+      V2InlineCliReference$zodSchema
+    ])).nullable().optional(),
+    judges: array(string2()).nullable().optional().describe("Replacement success criteria for the eval."),
+    mcps: array(union([
+      V2InlineMcpDefinition$zodSchema,
+      V2SavedMcpReference$zodSchema
+    ])).nullable().optional(),
     name: string2().optional(),
-    prompt: string2().optional(),
-    skill_id: string2().optional(),
-    type: UpdateV2EvalRequestType$zodSchema.optional()
+    persona: string2().optional().describe("Replacement eval persona. Pass null to clear it; omit it to leave the saved value unchanged."),
+    prompts: array(string2()).nullable().optional(),
+    runs: int().optional(),
+    tags: array(string2()).nullable().optional().describe("Replacement labels for the eval. Pass an empty array to clear all tags.")
   });
 });
 
@@ -57399,7 +57514,7 @@ var init_evalsV2EvalsUpdate2 = __esm(() => {
     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,
@@ -62082,7 +62197,7 @@ function createMCPServer(deps) {
       sizes: ["any"]
     }],
     name: "LightsageMcp",
-    version: "0.3.0"
+    version: "0.3.1"
   });
   const getClient = deps.getSDK || (() => new LightsageMcpCore({
     security: deps.security,
@@ -62379,21 +62494,23 @@ Returns one audited page with its score, priority, supporting evidence, and reco
       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.`
     },
     {
       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
+
+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.`
     },
     {
       name: "evals-v2-evals-retrieve",
       description: `Retrieve Eval
 
-Returns one eval and its judge configuration.`
+Returns 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",
@@ -62433,7 +62550,7 @@ Permanently deletes an eval configuration and returns its last saved state.`
     },
     {
       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",
@@ -63431,7 +63548,7 @@ 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;">
             \uD83D\uDCE5 Download MCP Bundle
           </a>
         </div>
@@ -66309,7 +66426,7 @@ var routes = buildRouteMap({
 var app = buildApplication(routes, {
   name: "mcp",
   versionInfo: {
-    currentVersion: "0.3.0"
+    currentVersion: "0.3.1"
   }
 });
 run(app, process3.argv.slice(2), buildContext(process3));
@@ -66317,5 +66434,5 @@ export {
   app
 };
 
-//# debugId=E8194E142237D9EF64756E2164756E21
+//# debugId=D144B9026D7DE38264756E2164756E21
 //# sourceMappingURL=mcp-server.js.map