@lightsage/mcp 0.2.2 → 0.3.1

package/esm/tool-names.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"tool-names.d.ts","sourceRoot":"","sources":["../src/tool-names.ts"],"names":[],"mappings":"AACA,eAAO,MAAM,SAAS,EAAE,KAAK,CAAC;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,WAAW,EAAE,MAAM,CAAA;CAAE,CAqNlE,CAAC"}
+{"version":3,"file":"tool-names.d.ts","sourceRoot":"","sources":["../src/tool-names.ts"],"names":[],"mappings":"AACA,eAAO,MAAM,SAAS,EAAE,KAAK,CAAC;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,WAAW,EAAE,MAAM,CAAA;CAAE,CAyNlE,CAAC"}

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 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, 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",
@@ -150,11 +150,15 @@ 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-monitor",
-        "description": "Monitor Runs\n\nStreams an immediate snapshot followed by live status and sanitized activity. Provide `run_id` to monitor one public v2 eval run, or omit it to capture all active runs when the connection opens."
+        "name": "evals-analysis",
+        "description": "Retrieve eval run analysis\n\nReturns the persisted analysis produced for each completed attempt in one eval run. The analysis includes the verdict, outcome explanation, findings, bottleneck, and analysis sections. Active runs return the analysis available so far together with current run status and progress."
+    },
+    {
+        "name": "evals-trace",
+        "description": "Retrieve an eval run trace\n\nReturns a finite page of sanitized execution events for one eval run. Use `next_after_id` as `after_id` to retrieve the next page. Use the monitor operation instead when following a run live."
     },
     {
         "name": "eval-runs-v2-eval-runs-list",
@@ -166,7 +170,7 @@ export const toolNames = [
     },
     {
         "name": "eval-results-v2-eval-results-list",
-        "description": "List Results\n\nReturns completed custom-eval results from onboarding, v1, and v2 run flows, ordered newest first. Result IDs are different from eval run IDs."
+        "description": "List Results\n\nReturns completed custom-eval results from all supported run flows, ordered newest first. Result IDs are different from eval run IDs."
     },
     {
         "name": "eval-results-v2-eval-results-retrieve",

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,oHAAoH;KACpI;IACD;QACE,MAAM,EAAE,yBAAyB;QACjC,aAAa,EAAE,gEAAgE;KAChF;IACD;QACE,MAAM,EAAE,uBAAuB;QAC/B,aAAa,EAAE,yGAAyG;KACzH;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,eAAe;QACvB,aAAa,EAAE,oNAAoN;KACpO;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,gKAAgK;KAChL;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": "0.2.2",
+  "version": "0.3.1",
   "author": "Speakeasy",
   "type": "module",
   "sideEffects": false,

package/src/funcs/evalResultsV2EvalResultsList.ts

@@ -30,7 +30,7 @@ import { Result } from "../types/fp.js";
  * List Results
  *
  * @remarks
- * Returns completed custom-eval results from onboarding, v1, and v2 run flows, ordered newest first. Result IDs are different from eval run IDs.
+ * Returns completed custom-eval results from all supported run flows, ordered newest first. Result IDs are different from eval run IDs.
  */
 export function evalResultsV2EvalResultsList(
   client$: LightsageMcpCore,

package/src/funcs/evalsAnalysis.ts

@@ -0,0 +1,153 @@
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ * @generated-id: cb2b198fcbdb
+ */
+
+import { LightsageMcpCore } from "../core.js";
+import { encodeSimple } from "../lib/encodings.js";
+import { compactMap } from "../lib/primitives.js";
+import { safeParse } from "../lib/schemas.js";
+import { RequestOptions } from "../lib/sdks.js";
+import { extractSecurity, resolveGlobalSecurity } from "../lib/security.js";
+import { pathToFunc } from "../lib/url.js";
+import { APIError } from "../models/errors/apierror.js";
+import {
+  ConnectionError,
+  InvalidRequestError,
+  RequestAbortedError,
+  RequestTimeoutError,
+  UnexpectedClientError,
+} from "../models/errors/httpclienterrors.js";
+import { SDKValidationError } from "../models/errors/sdkvalidationerror.js";
+import {
+  V2EvalRunsAnalysisRequest,
+  V2EvalRunsAnalysisRequest$zodSchema,
+} from "../models/v2evalrunsanalysisop.js";
+import { APICall, APIPromise } from "../types/async.js";
+import { Result } from "../types/fp.js";
+
+/**
+ * Retrieve eval run analysis
+ *
+ * @remarks
+ * Returns the persisted analysis produced for each completed attempt in one eval run. The analysis includes the verdict, outcome explanation, findings, bottleneck, and analysis sections. Active runs return the analysis available so far together with current run status and progress.
+ */
+export function evalsAnalysis(
+  client$: LightsageMcpCore,
+  request: V2EvalRunsAnalysisRequest,
+  options?: RequestOptions,
+): APIPromise<
+  Result<
+    Response,
+    | APIError
+    | SDKValidationError
+    | UnexpectedClientError
+    | InvalidRequestError
+    | RequestAbortedError
+    | RequestTimeoutError
+    | ConnectionError
+  >
+> {
+  return new APIPromise($do(
+    client$,
+    request,
+    options,
+  ));
+}
+
+async function $do(
+  client$: LightsageMcpCore,
+  request: V2EvalRunsAnalysisRequest,
+  options?: RequestOptions,
+): Promise<
+  [
+    Result<
+      Response,
+      | APIError
+      | SDKValidationError
+      | UnexpectedClientError
+      | InvalidRequestError
+      | RequestAbortedError
+      | RequestTimeoutError
+      | ConnectionError
+    >,
+    APICall,
+  ]
+> {
+  const parsed$ = safeParse(
+    request,
+    (value$) => V2EvalRunsAnalysisRequest$zodSchema.parse(value$),
+    "Input validation failed",
+  );
+  if (!parsed$.ok) {
+    return [parsed$, { status: "invalid" }];
+  }
+  const payload$ = parsed$.value;
+  const body$ = null;
+
+  const pathParams$ = {
+    run_id: encodeSimple("run_id", payload$.run_id, {
+      explode: false,
+      charEncoding: "percent",
+    }),
+  };
+  const path$ = pathToFunc("/v2/eval-runs/{run_id}/analysis")(
+    pathParams$,
+  );
+
+  const headers$ = new Headers(compactMap({
+    Accept: "application/json",
+  }));
+  const securityInput = await extractSecurity(client$._options.security);
+  const requestSecurity = resolveGlobalSecurity(securityInput);
+
+  const context = {
+    options: client$._options,
+    baseURL: options?.serverURL ?? client$._baseURL ?? "",
+    operationID: "v2_eval_runs_analysis",
+    oAuth2Scopes: null,
+    resolvedSecurity: requestSecurity,
+    securitySource: client$._options.security,
+    retryConfig: options?.retries
+      || client$._options.retryConfig
+      || { strategy: "none" },
+    retryCodes: options?.retryCodes || [
+      "429",
+      "500",
+      "502",
+      "503",
+      "504",
+    ],
+  };
+
+  const requestRes = client$._createRequest(context, {
+    security: requestSecurity,
+    method: "GET",
+    baseURL: options?.serverURL,
+    path: path$,
+    headers: headers$,
+    body: body$,
+    userAgent: client$._options.userAgent,
+    timeoutMs: options?.timeoutMs || client$._options.timeoutMs
+      || -1,
+  }, options);
+  if (!requestRes.ok) {
+    return [requestRes, { status: "invalid" }];
+  }
+  const req$ = requestRes.value;
+
+  const doResult = await client$._do(req$, {
+    context,
+    errorCodes: [],
+    retryConfig: context.retryConfig,
+    retryCodes: context.retryCodes,
+  });
+  if (!doResult.ok) {
+    return [doResult, { status: "request-error", request: req$ }];
+  }
+  return [doResult, {
+    status: "complete",
+    "request": req$,
+    response: doResult.value,
+  }];
+}

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/{evalsMonitor.ts → evalsTrace.ts}

@@ -1,10 +1,10 @@
 /*
  * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
- * @generated-id: da8fcf6d681f
+ * @generated-id: 696bcb18693c
  */
 
 import { LightsageMcpCore } from "../core.js";
-import { encodeFormQuery } from "../lib/encodings.js";
+import { encodeFormQuery, encodeSimple } from "../lib/encodings.js";
 import { compactMap } from "../lib/primitives.js";
 import { safeParse } from "../lib/schemas.js";
 import { RequestOptions } from "../lib/sdks.js";
@@ -20,26 +20,21 @@ import {
 } from "../models/errors/httpclienterrors.js";
 import { SDKValidationError } from "../models/errors/sdkvalidationerror.js";
 import {
-  V2EvalRunsMonitorRequest,
-  V2EvalRunsMonitorRequest$zodSchema,
-} from "../models/v2evalrunsmonitorop.js";
+  V2EvalRunsTraceRequest,
+  V2EvalRunsTraceRequest$zodSchema,
+} from "../models/v2evalrunstraceop.js";
 import { APICall, APIPromise } from "../types/async.js";
 import { Result } from "../types/fp.js";
 
-export enum MonitorAcceptEnum {
-  applicationJsonAccept = "application/json",
-  textEventStreamAccept = "text/event-stream",
-}
-
 /**
- * Monitor Runs
+ * Retrieve an eval run trace
  *
  * @remarks
- * Streams an immediate snapshot followed by live status and sanitized activity. Provide `run_id` to monitor one public v2 eval run, or omit it to capture all active runs when the connection opens.
+ * Returns a finite page of sanitized execution events for one eval run. Use `next_after_id` as `after_id` to retrieve the next page. Use the monitor operation instead when following a run live.
  */
-export function evalsMonitor(
+export function evalsTrace(
   client$: LightsageMcpCore,
-  request?: V2EvalRunsMonitorRequest | undefined,
+  request: V2EvalRunsTraceRequest,
   options?: RequestOptions,
 ): APIPromise<
   Result<
@@ -62,8 +57,8 @@ export function evalsMonitor(
 
 async function $do(
   client$: LightsageMcpCore,
-  request?: V2EvalRunsMonitorRequest | undefined,
-  options?: RequestOptions & { acceptHeaderOverride?: MonitorAcceptEnum },
+  request: V2EvalRunsTraceRequest,
+  options?: RequestOptions,
 ): Promise<
   [
     Result<
@@ -81,7 +76,7 @@ async function $do(
 > {
   const parsed$ = safeParse(
     request,
-    (value$) => V2EvalRunsMonitorRequest$zodSchema.optional().parse(value$),
+    (value$) => V2EvalRunsTraceRequest$zodSchema.parse(value$),
     "Input validation failed",
   );
   if (!parsed$.ok) {
@@ -89,16 +84,23 @@ async function $do(
   }
   const payload$ = parsed$.value;
   const body$ = null;
-  const path$ = pathToFunc("/v2/eval-runs/monitor")();
+
+  const pathParams$ = {
+    run_id: encodeSimple("run_id", payload$.run_id, {
+      explode: false,
+      charEncoding: "percent",
+    }),
+  };
+  const path$ = pathToFunc("/v2/eval-runs/{run_id}/trace")(
+    pathParams$,
+  );
   const query$ = encodeFormQuery({
-    "after_id": payload$?.after_id,
-    "detail": payload$?.detail,
-    "run_id": payload$?.run_id,
+    "after_id": payload$.after_id,
+    "limit": payload$.limit,
   });
 
   const headers$ = new Headers(compactMap({
-    Accept: options?.acceptHeaderOverride
-      || "application/json;q=1, text/event-stream;q=0",
+    Accept: "application/json",
   }));
   const securityInput = await extractSecurity(client$._options.security);
   const requestSecurity = resolveGlobalSecurity(securityInput);
@@ -106,7 +108,7 @@ async function $do(
   const context = {
     options: client$._options,
     baseURL: options?.serverURL ?? client$._baseURL ?? "",
-    operationID: "v2_eval_runs_monitor",
+    operationID: "v2_eval_runs_trace",
     oAuth2Scopes: null,
     resolvedSecurity: requestSecurity,
     securitySource: client$._options.security,

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 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, 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/v0.2.2/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: "0.2.2",
-  genVersion: "2.935.1",
-  userAgent: "speakeasy-sdk/mcp-typescript 0.2.2 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: "0.2.2",
+    currentVersion: "0.3.1",
   },
 });
 

package/src/mcp-server/server.ts

@@ -28,8 +28,9 @@ import { tool$evalResultsV2EvalResultsList } from "./tools/evalResultsV2EvalResu
 import { tool$evalResultsV2EvalResultsRetrieve } from "./tools/evalResultsV2EvalResultsRetrieve.js";
 import { tool$evalRunsV2EvalRunsList } from "./tools/evalRunsV2EvalRunsList.js";
 import { tool$evalRunsV2EvalRunsRetrieve } from "./tools/evalRunsV2EvalRunsRetrieve.js";
-import { tool$evalsMonitor } from "./tools/evalsMonitor.js";
+import { tool$evalsAnalysis } from "./tools/evalsAnalysis.js";
 import { tool$evalsRun } from "./tools/evalsRun.js";
+import { tool$evalsTrace } from "./tools/evalsTrace.js";
 import { tool$evalsV2EvalConfigurationsCreate } from "./tools/evalsV2EvalConfigurationsCreate.js";
 import { tool$evalsV2EvalConfigurationsDelete } from "./tools/evalsV2EvalConfigurationsDelete.js";
 import { tool$evalsV2EvalConfigurationsList } from "./tools/evalsV2EvalConfigurationsList.js";
@@ -91,7 +92,7 @@ export function createMCPServer(deps: {
       sizes: ["any"],
     }],
     name: "LightsageMcp",
-    version: "0.2.2",
+    version: "0.3.1",
   });
 
   const getClient = deps.getSDK || (() =>
@@ -174,7 +175,8 @@ export function createMCPServer(deps: {
   tool(tool$evalsV2EvalConfigurationsUpdate);
   tool(tool$evalsV2EvalConfigurationsDelete);
   tool(tool$evalsRun);
-  tool(tool$evalsMonitor);
+  tool(tool$evalsAnalysis);
+  tool(tool$evalsTrace);
   tool(tool$evalRunsV2EvalRunsList);
   tool(tool$evalRunsV2EvalRunsRetrieve);
   tool(tool$evalResultsV2EvalResultsList);

package/src/mcp-server/tools/evalResultsV2EvalResultsList.ts

@@ -15,7 +15,7 @@ export const tool$evalResultsV2EvalResultsList: ToolDefinition<typeof args> = {
   name: "eval-results-v2-eval-results-list",
   description: `List Results
 
-Returns completed custom-eval results from onboarding, v1, and v2 run flows, ordered newest first. Result IDs are different from eval run IDs.`,
+Returns completed custom-eval results from all supported run flows, ordered newest first. Result IDs are different from eval run IDs.`,
   annotations: {
     "title": "",
     "destructiveHint": false,

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

@@ -0,0 +1,43 @@
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ * @generated-id: 323a92ae7245
+ */
+
+import { evalsAnalysis } from "../../funcs/evalsAnalysis.js";
+import { V2EvalRunsAnalysisRequest$zodSchema } from "../../models/v2evalrunsanalysisop.js";
+import { formatResult, ToolDefinition } from "../tools.js";
+
+const args = {
+  request: V2EvalRunsAnalysisRequest$zodSchema,
+};
+
+export const tool$evalsAnalysis: ToolDefinition<typeof args> = {
+  name: "evals-analysis",
+  description: `Retrieve eval run analysis
+
+Returns the persisted analysis produced for each completed attempt in one eval run. The analysis includes the verdict, outcome explanation, findings, bottleneck, and analysis sections. Active runs return the analysis available so far together with current run status and progress.`,
+  annotations: {
+    "title": "",
+    "destructiveHint": false,
+    "idempotentHint": false,
+    "openWorldHint": false,
+    "readOnlyHint": true,
+  },
+  args,
+  tool: async (client, args, ctx) => {
+    const [result] = await evalsAnalysis(
+      client,
+      args.request,
+      { fetchOptions: { signal: ctx.signal } },
+    ).$inspect();
+
+    if (!result.ok) {
+      return {
+        content: [{ type: "text", text: result.error.message }],
+        isError: true,
+      };
+    }
+
+    return formatResult(result.value);
+  },
+};

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/{evalsMonitor.ts → evalsTrace.ts}

@@ -1,21 +1,21 @@
 /*
  * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
- * @generated-id: f7c3344a66fd
+ * @generated-id: d4ce799d0d36
  */
 
-import { evalsMonitor } from "../../funcs/evalsMonitor.js";
-import { V2EvalRunsMonitorRequest$zodSchema } from "../../models/v2evalrunsmonitorop.js";
+import { evalsTrace } from "../../funcs/evalsTrace.js";
+import { V2EvalRunsTraceRequest$zodSchema } from "../../models/v2evalrunstraceop.js";
 import { formatResult, ToolDefinition } from "../tools.js";
 
 const args = {
-  request: V2EvalRunsMonitorRequest$zodSchema.optional(),
+  request: V2EvalRunsTraceRequest$zodSchema,
 };
 
-export const tool$evalsMonitor: ToolDefinition<typeof args> = {
-  name: "evals-monitor",
-  description: `Monitor Runs
+export const tool$evalsTrace: ToolDefinition<typeof args> = {
+  name: "evals-trace",
+  description: `Retrieve an eval run trace
 
-Streams an immediate snapshot followed by live status and sanitized activity. Provide \`run_id\` to monitor one public v2 eval run, or omit it to capture all active runs when the connection opens.`,
+Returns a finite page of sanitized execution events for one eval run. Use \`next_after_id\` as \`after_id\` to retrieve the next page. Use the monitor operation instead when following a run live.`,
   annotations: {
     "title": "",
     "destructiveHint": false,
@@ -25,7 +25,7 @@ Streams an immediate snapshot followed by live status and sanitized activity. Pr
   },
   args,
   tool: async (client, args, ctx) => {
-    const [result] = await evalsMonitor(
+    const [result] = await evalsTrace(
       client,
       args.request,
       { fetchOptions: { signal: ctx.signal } },

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 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, 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,