@vellumai/assistant 0.6.0 → 0.6.2

package/src/runtime/routes/conversation-analysis-routes.ts

@@ -0,0 +1,185 @@
+/**
+ * Route handler for conversation analysis.
+ *
+ * POST /v1/conversations/:id/analyze — analyze a conversation via a new
+ * agent loop that produces a structured self-assessment.
+ */
+
+import type { ServerMessage } from "../../daemon/message-protocol.js";
+import {
+  addMessage,
+  createConversation,
+  getConversation,
+  getMessages,
+} from "../../memory/conversation-crud.js";
+import { resolveConversationId } from "../../memory/conversation-key-store.js";
+import { getLogger } from "../../util/logger.js";
+import { buildAssistantEvent } from "../assistant-event.js";
+import { DAEMON_INTERNAL_ASSISTANT_ID } from "../assistant-scope.js";
+import { httpError } from "../http-errors.js";
+import type { RouteDefinition } from "../http-router.js";
+import type { SendMessageDeps } from "../http-types.js";
+
+const log = getLogger("conversation-analysis-routes");
+
+// ---------------------------------------------------------------------------
+// Dependency types — injected by the daemon at wiring time
+// ---------------------------------------------------------------------------
+
+export interface ConversationAnalysisDeps {
+  sendMessageDeps: SendMessageDeps;
+  buildConversationDetailResponse: (
+    id: string,
+  ) => Record<string, unknown> | null;
+}
+
+// ---------------------------------------------------------------------------
+// Route definitions
+// ---------------------------------------------------------------------------
+
+export function conversationAnalysisRouteDefinitions(
+  deps: ConversationAnalysisDeps,
+): RouteDefinition[] {
+  return [
+    {
+      endpoint: "conversations/:id/analyze",
+      method: "POST",
+      policyKey: "conversations/analyze",
+      summary: "Analyze a conversation",
+      description:
+        "Create a new conversation with a structured self-assessment of an existing conversation.",
+      tags: ["conversations"],
+      handler: async ({ params }) => {
+        // a. Resolve conversation ID
+        const resolvedId = resolveConversationId(params.id);
+        if (!resolvedId) {
+          return httpError(
+            "NOT_FOUND",
+            `Conversation ${params.id} not found`,
+            404,
+          );
+        }
+
+        // b. Load the conversation
+        const conversation = getConversation(resolvedId);
+        if (!conversation) {
+          return httpError(
+            "NOT_FOUND",
+            `Conversation ${resolvedId} not found`,
+            404,
+          );
+        }
+
+        // c. Reject private conversations
+        if (conversation.conversationType === "private") {
+          return httpError(
+            "FORBIDDEN",
+            "Private conversations cannot be analyzed",
+            403,
+          );
+        }
+
+        // d. Check for messages
+        const existingMessages = getMessages(resolvedId);
+        if (existingMessages.length === 0) {
+          return httpError(
+            "BAD_REQUEST",
+            "Conversation has no messages to analyze",
+            400,
+          );
+        }
+
+        // e. Build the analysis transcript
+        const { buildAnalysisTranscript } =
+          await import("../../export/transcript-formatter.js");
+        const transcript = buildAnalysisTranscript(resolvedId);
+
+        // f. Create a new conversation for the analysis
+        const newConv = createConversation({
+          title: `Analysis: ${conversation.title ?? "Untitled"}`,
+        });
+
+        // g. Build the analysis prompt
+        const prompt = `<transcript>
+${transcript}
+</transcript>
+
+Analyze the conversation above. Provide a structured self-assessment:
+
+1. **Summary**: What was the user trying to accomplish? What was the outcome?
+2. **What went well**: Effective tool usage, good reasoning, helpful responses, problem-solving patterns.
+3. **What went wrong**: Errors, unnecessary tool calls, incorrect assumptions, wasted turns, misunderstandings.
+4. **Root causes**: Why did failures happen? Missing context? Wrong approach? Tool limitations?
+5. **Recommendations**: Specific, actionable improvements for similar conversations next time.
+
+Be honest and specific. Reference particular moments in the transcript. Focus on patterns that generalize beyond this specific conversation.
+
+Do not use tools during analysis. If you identify insights worth remembering for future conversations, include them in the response as explicit memory candidates instead of saving them directly.`;
+
+        // h. Persist the user message
+        const message = await addMessage(
+          newConv.id,
+          "user",
+          JSON.stringify([{ type: "text", text: prompt }]),
+          { provenanceTrustClass: "unknown" as const },
+        );
+        const messageId = message.id;
+
+        // i. Load the conversation into memory with untrusted analysis context
+        const analysisConversation =
+          await deps.sendMessageDeps.getOrCreateConversation(newConv.id);
+        analysisConversation.setTrustContext({
+          trustClass: "unknown",
+          sourceChannel: "vellum",
+        });
+        await analysisConversation.ensureActorScopedHistory();
+        // Analysis runs over attacker-influenced transcript content, so do not
+        // expose any tools, even when a live client is available.
+        analysisConversation.setSubagentAllowedTools(new Set<string>());
+
+        const hasLiveSubscriber =
+          deps.sendMessageDeps.assistantEventHub.hasSubscribersForEvent({
+            assistantId: DAEMON_INTERNAL_ASSISTANT_ID,
+            conversationId: newConv.id,
+          });
+
+        // j. Build onEvent using inline hub publisher
+        const onEvent = (msg: ServerMessage) => {
+          deps.sendMessageDeps.assistantEventHub.publish(
+            buildAssistantEvent(DAEMON_INTERNAL_ASSISTANT_ID, msg, newConv.id),
+          );
+        };
+        analysisConversation.updateClient(onEvent, !hasLiveSubscriber);
+
+        // k. Set up processing state (required by runAgentLoop guard)
+        analysisConversation.processing = true;
+        analysisConversation.abortController = new AbortController();
+        analysisConversation.currentRequestId = crypto.randomUUID();
+
+        // l. Fire-and-forget the agent loop
+        analysisConversation
+          .runAgentLoop(prompt, messageId, onEvent, {
+            isInteractive: hasLiveSubscriber,
+            isUserMessage: true,
+          })
+          .catch((err) => {
+            log.error(
+              { err, conversationId: newConv.id },
+              "Analysis agent loop failed",
+            );
+          });
+
+        // m. Return the new conversation detail
+        const detail = deps.buildConversationDetailResponse(newConv.id);
+        if (!detail) {
+          return httpError(
+            "INTERNAL_ERROR",
+            `Analysis conversation ${newConv.id} could not be loaded`,
+            500,
+          );
+        }
+        return Response.json(detail);
+      },
+    },
+  ];
+}

package/src/runtime/routes/conversation-management-routes.ts

@@ -344,12 +344,6 @@ export function conversationManagementRouteDefinitions(
             targetId: segId,
           });
         }
-        for (const itemId of result.orphanedItemIds) {
-          enqueueMemoryJob("delete_qdrant_vectors", {
-            targetType: "item",
-            targetId: itemId,
-          });
-        }
         for (const summaryId of result.deletedSummaryIds) {
           enqueueMemoryJob("delete_qdrant_vectors", {
             targetType: "summary",
@@ -359,7 +353,6 @@ export function conversationManagementRouteDefinitions(
         log.info(
           {
             conversationId: resolvedId,
-            unsuperseded: result.unsupersededItemIds.length,
             summariesDeleted: result.deletedSummaryIds.length,
             jobsCancelled: result.cancelledJobCount,
           },
@@ -367,7 +360,7 @@ export function conversationManagementRouteDefinitions(
         );
         return Response.json({
           wiped: true,
-          unsupersededItems: result.unsupersededItemIds.length,
+          unsupersededItems: 0,
           deletedSummaries: result.deletedSummaryIds.length,
           cancelledJobs: result.cancelledJobCount,
         });
@@ -417,12 +410,6 @@ export function conversationManagementRouteDefinitions(
             targetId: segId,
           });
         }
-        for (const itemId of deleted.orphanedItemIds) {
-          enqueueMemoryJob("delete_qdrant_vectors", {
-            targetType: "item",
-            targetId: itemId,
-          });
-        }
         for (const summaryId of deleted.deletedSummaryIds) {
           enqueueMemoryJob("delete_qdrant_vectors", {
             targetType: "summary",

package/src/runtime/routes/conversation-query-routes.ts

@@ -43,7 +43,10 @@ import {
 } from "../../daemon/handlers/conversation-history.js";
 import { deleteQueuedMessage } from "../../daemon/handlers/conversations.js";
 import { getAssistantMessageIdsInTurn } from "../../memory/conversation-crud.js";
-import { getRequestLogsByMessageId } from "../../memory/llm-request-log-store.js";
+import {
+  getRequestLogById,
+  getRequestLogsByMessageId,
+} from "../../memory/llm-request-log-store.js";
 import { getMemoryRecallLogByMessageIds } from "../../memory/memory-recall-log-store.js";
 import { resolvePricingForUsage } from "../../util/pricing.js";
 import { httpError } from "../http-errors.js";
@@ -487,8 +490,8 @@ export function conversationQueryRouteDefinitions(
             );
             return {
               id: log.id,
-              requestPayload,
-              responsePayload,
+              requestPayload: null,
+              responsePayload: null,
               createdAt: log.createdAt,
               ...result,
             };
@@ -498,6 +501,49 @@ export function conversationQueryRouteDefinitions(
       },
     },
 
+    // ── Raw payload for a single LLM request log ─────────────────────
+    {
+      endpoint: "llm-request-logs/:id/payload",
+      method: "GET",
+      policyKey: "llm-request-logs/payload",
+      summary: "Get raw payload for a single LLM request log",
+      description:
+        "Return the full request and response payloads for a specific log entry.",
+      tags: ["messages"],
+      responseBody: z.object({
+        id: z.string(),
+        requestPayload: z.unknown(),
+        responsePayload: z.unknown(),
+      }),
+      handler: ({ params }) => {
+        const logId = params.id;
+        if (!logId) {
+          return httpError("BAD_REQUEST", "log id is required", 400);
+        }
+        const log = getRequestLogById(logId);
+        if (!log) {
+          return httpError("NOT_FOUND", "log not found", 404);
+        }
+        let requestPayload: unknown;
+        try {
+          requestPayload = JSON.parse(log.requestPayload);
+        } catch {
+          requestPayload = log.requestPayload;
+        }
+        let responsePayload: unknown;
+        try {
+          responsePayload = JSON.parse(log.responsePayload);
+        } catch {
+          responsePayload = log.responsePayload;
+        }
+        return Response.json({
+          id: log.id,
+          requestPayload,
+          responsePayload,
+        });
+      },
+    },
+
     // ── Delete queued message ─────────────────────────────────────────
     {
       endpoint: "messages/queued/:id",