Python package improvements

Added an endpoint for getting the actual stored responses, and used it to test and improve the python package.

app/.env.example

@@ -32,5 +32,5 @@ NEXT_PUBLIC_HOST="http://localhost:3000"
 GITHUB_CLIENT_ID="your_client_id"
 GITHUB_CLIENT_SECRET="your_secret"
 
-OPENPIPE_BASE_URL="http://localhost:3000/api"
+OPENPIPE_BASE_URL="http://localhost:3000/api/v1"
 OPENPIPE_API_KEY="your_key"

app/@types/nextjs-routes.d.ts

@@ -12,12 +12,11 @@ declare module "nextjs-routes" {
 
   export type Route =
     | StaticRoute<"/account/signin">
-    | DynamicRoute<"/api/[...trpc]", { "trpc": string[] }>
     | DynamicRoute<"/api/auth/[...nextauth]", { "nextauth": string[] }>
     | StaticRoute<"/api/experiments/og-image">
-    | StaticRoute<"/api/openapi">
-    | StaticRoute<"/api/sentry-example-api">
     | DynamicRoute<"/api/trpc/[trpc]", { "trpc": string }>
+    | DynamicRoute<"/api/v1/[...trpc]", { "trpc": string[] }>
+    | StaticRoute<"/api/v1/openapi">
     | StaticRoute<"/dashboard">
     | DynamicRoute<"/data/[id]", { "id": string }>
     | StaticRoute<"/data">

app/prisma/schema.prisma

@@ -112,17 +112,17 @@ model ScenarioVariantCell {
 model ModelResponse {
     id String @id @default(uuid()) @db.Uuid
 
-    cacheKey        String
-    requestedAt      DateTime?
-    receivedAt       DateTime?
-    respPayload      Json?
-    cost             Float?
-    inputTokens      Int?
-    outputTokens     Int?
-    statusCode       Int?
-    errorMessage     String?
-    retryTime        DateTime?
-    outdated         Boolean   @default(false)
+    cacheKey     String
+    requestedAt  DateTime?
+    receivedAt   DateTime?
+    respPayload  Json?
+    cost         Float?
+    inputTokens  Int?
+    outputTokens Int?
+    statusCode   Int?
+    errorMessage String?
+    retryTime    DateTime?
+    outdated     Boolean   @default(false)
 
     createdAt DateTime @default(now())
     updatedAt DateTime @updatedAt
@@ -273,8 +273,8 @@ model LoggedCall {
     projectId String   @db.Uuid
     project   Project? @relation(fields: [projectId], references: [id], onDelete: Cascade)
 
-    model        String?
-    tags         LoggedCallTag[]
+    model String?
+    tags  LoggedCallTag[]
 
     createdAt DateTime @default(now())
     updatedAt DateTime @updatedAt
@@ -295,7 +295,7 @@ model LoggedCallModelResponse {
     errorMessage String?
 
     requestedAt DateTime
-    receivedAt   DateTime
+    receivedAt  DateTime
 
     // Note: the function to calculate the cacheKey should include the project
     // ID so we don't share cached responses between projects, which could be an
@@ -340,8 +340,8 @@ model ApiKey {
     name   String
     apiKey String @unique
 
-    projectId String   @db.Uuid
-    project   Project? @relation(fields: [projectId], references: [id], onDelete: Cascade)
+    projectId String  @db.Uuid
+    project   Project @relation(fields: [projectId], references: [id], onDelete: Cascade)
 
     createdAt DateTime @default(now())
     updatedAt DateTime @updatedAt

app/prisma/seed.ts

@@ -2,6 +2,7 @@ import { prisma } from "~/server/db";
 import dedent from "dedent";
 import { generateNewCell } from "~/server/utils/generateNewCell";
 import { promptConstructorVersion } from "~/promptConstructor/version";
+import { env } from "~/env.mjs";
 
 const defaultId = "11111111-1111-1111-1111-111111111111";
 
@@ -16,6 +17,16 @@ const project =
     data: { id: defaultId },
   }));
 
+if (env.OPENPIPE_API_KEY) {
+  await prisma.apiKey.create({
+    data: {
+      projectId: project.id,
+      name: "Default API Key",
+      apiKey: env.OPENPIPE_API_KEY,
+    },
+  });
+}
+
 await prisma.experiment.deleteMany({
   where: {
     id: defaultId,

app/src/components/OutputsTable/OutputCell/OutputCell.tsx

@@ -33,7 +33,7 @@ export default function OutputCell({
 
   if (!templateHasVariables) disabledReason = "Add a value to the scenario variables to see output";
 
-  const [refetchInterval, setRefetchInterval] = useState(0);
+  const [refetchInterval, setRefetchInterval] = useState<number | false>(false);
   const { data: cell, isLoading: queryLoading } = api.scenarioVariantCells.get.useQuery(
     { scenarioId: scenario.id, variantId: variant.id },
     { refetchInterval },
@@ -64,7 +64,8 @@ export default function OutputCell({
     cell.retrievalStatus === "PENDING" ||
     cell.retrievalStatus === "IN_PROGRESS" ||
     hardRefetching;
-  useEffect(() => setRefetchInterval(awaitingOutput ? 1000 : 0), [awaitingOutput]);
+
+  useEffect(() => setRefetchInterval(awaitingOutput ? 1000 : false), [awaitingOutput]);
 
   // TODO: disconnect from socket if we're not streaming anymore
   const streamedMessage = useSocket<OutputSchema>(cell?.id);
@@ -120,8 +121,13 @@ export default function OutputCell({
             ? response.receivedAt.getTime()
             : Date.now();
           if (response.requestedAt) {
-            numWaitingMessages = Math.floor(
-              (relativeWaitingTime - response.requestedAt.getTime()) / WAITING_MESSAGE_INTERVAL,
+            numWaitingMessages = Math.min(
+              Math.floor(
+                (relativeWaitingTime - response.requestedAt.getTime()) / WAITING_MESSAGE_INTERVAL,
+              ),
+              // Don't try to render more than 15, it'll use too much CPU and
+              // break the page
+              15,
             );
           }
           return (

app/src/components/OutputsTable/ScenarioRow.tsx

@@ -10,6 +10,7 @@ const ScenarioRow = (props: {
   variants: PromptVariant[];
   canHide: boolean;
   rowStart: number;
+  isFirst: boolean;
   isLast: boolean;
 }) => {
   const [isHovered, setIsHovered] = useState(false);
@@ -23,11 +24,13 @@ const ScenarioRow = (props: {
         onMouseLeave={() => setIsHovered(false)}
         sx={isHovered ? highlightStyle : undefined}
         bgColor="white"
-        borderLeftWidth={1}
-        {...borders}
         rowStart={props.rowStart}
         colStart={1}
+        borderLeftWidth={1}
+        borderTopWidth={props.isFirst ? 1 : 0}
+        borderTopLeftRadius={props.isFirst ? 8 : 0}
         borderBottomLeftRadius={props.isLast ? 8 : 0}
+        {...borders}
       >
         <ScenarioEditor scenario={props.scenario} hovered={isHovered} canHide={props.canHide} />
       </GridItem>
@@ -40,6 +43,8 @@ const ScenarioRow = (props: {
           bgColor="white"
           rowStart={props.rowStart}
           colStart={i + 2}
+          borderTopWidth={props.isFirst ? 1 : 0}
+          borderTopRightRadius={props.isFirst && i === props.variants.length - 1 ? 8 : 0}
           borderBottomRightRadius={props.isLast && i === props.variants.length - 1 ? 8 : 0}
           {...borders}
         >

app/src/components/OutputsTable/ScenariosHeader.tsx

@@ -48,20 +48,7 @@ export const ScenariosHeader = () => {
   );
 
   return (
-    <HStack
-      w="100%"
-      py={cellPadding.y}
-      px={cellPadding.x}
-      align="center"
-      spacing={0}
-      borderTopRightRadius={8}
-      borderTopLeftRadius={8}
-      bgColor="white"
-      borderWidth={1}
-      borderBottomWidth={0}
-      borderColor="gray.300"
-      mt={8}
-    >
+    <HStack w="100%" py={cellPadding.y} px={cellPadding.x} align="center" spacing={0}>
       <Text fontSize={16} fontWeight="bold">
         Scenarios ({scenarios.data?.count})
       </Text>

app/src/components/OutputsTable/VariantStats.tsx

@@ -21,14 +21,18 @@ export default function VariantStats(props: { variant: PromptVariant }) {
         outputTokens: 0,
         scenarioCount: 0,
         outputCount: 0,
+        awaitingCompletions: false,
         awaitingEvals: false,
       },
       refetchInterval,
     },
   );
 
-  // Poll every two seconds while we are waiting for LLM retrievals to finish
-  useEffect(() => setRefetchInterval(data.awaitingEvals ? 5000 : 0), [data.awaitingEvals]);
+  // Poll every five seconds while we are waiting for LLM retrievals to finish
+  useEffect(
+    () => setRefetchInterval(data.awaitingCompletions || data.awaitingEvals ? 5000 : 0),
+    [data.awaitingCompletions, data.awaitingEvals],
+  );
 
   const [passColor, neutralColor, failColor] = useToken("colors", [
     "green.500",

app/src/components/OutputsTable/index.tsx

@@ -86,7 +86,6 @@ export default function OutputsTable({ experimentId }: { experimentId: string |
         colSpan={allCols - 1}
         rowStart={variantHeaderRows + 1}
         colStart={1}
-        {...borders}
         borderRightWidth={0}
       >
         <ScenariosHeader />
@@ -99,6 +98,7 @@ export default function OutputsTable({ experimentId }: { experimentId: string |
           scenario={scenario}
           variants={variants.data}
           canHide={visibleScenariosCount > 1}
+          isFirst={i === 0}
           isLast={i === visibleScenariosCount - 1}
         />
       ))}

app/src/modelProviders/openai-ChatCompletion/refinementActions.ts

@@ -120,9 +120,9 @@ export const refinementActions: Record<string, RefinementAction> = {
   "Convert to function call": {
     icon: TfiThought,
     description: "Use function calls to get output from the model in a more structured way.",
-    instructions: `OpenAI functions are a specialized way for an LLM to return output.
+    instructions: `OpenAI functions are a specialized way for an LLM to return its final output.
   
-      This is what a prompt looks like before adding a function:
+      Example 1 before:
   
       definePrompt("openai/ChatCompletion", {
         model: "gpt-4",
@@ -139,7 +139,7 @@ export const refinementActions: Record<string, RefinementAction> = {
         ],
       });
   
-      This is what one looks like after adding a function:
+      Example 1 after:
   
       definePrompt("openai/ChatCompletion", {
         model: "gpt-4",
@@ -156,7 +156,7 @@ export const refinementActions: Record<string, RefinementAction> = {
         ],
         functions: [
           {
-            name: "extract_sentiment",
+            name: "log_extracted_sentiment",
             parameters: {
               type: "object", // parameters must always be an object with a properties key
               properties: { // properties key is required
@@ -169,13 +169,13 @@ export const refinementActions: Record<string, RefinementAction> = {
           },
         ],
         function_call: {
-          name: "extract_sentiment",
+          name: "log_extracted_sentiment",
         },
       });
   
-      Here's another example of adding a function:
-  
-      Before:
+      =========
+
+      Example 2 before:
   
       definePrompt("openai/ChatCompletion", {
           model: "gpt-3.5-turbo",
@@ -197,7 +197,7 @@ export const refinementActions: Record<string, RefinementAction> = {
           temperature: 0,
       });
   
-      After:
+      Example 2 after:
   
       definePrompt("openai/ChatCompletion", {
           model: "gpt-3.5-turbo",
@@ -215,7 +215,7 @@ export const refinementActions: Record<string, RefinementAction> = {
           temperature: 0,
           functions: [
             {
-              name: "score_post",
+              name: "log_post_score",
               parameters: {
                 type: "object",
                 properties: {
@@ -227,13 +227,13 @@ export const refinementActions: Record<string, RefinementAction> = {
             },
           ],
           function_call: {
-            name: "score_post",
+            name: "log_post_score",
           },
         });
   
-      Another example
+      =========
   
-      Before:
+      Example 3 before:
   
       definePrompt("openai/ChatCompletion", {
         model: "gpt-3.5-turbo",
@@ -246,7 +246,7 @@ export const refinementActions: Record<string, RefinementAction> = {
         ],
       });
   
-      After:
+      Example 3 after:
   
       definePrompt("openai/ChatCompletion", {
         model: "gpt-3.5-turbo",
@@ -258,21 +258,24 @@ export const refinementActions: Record<string, RefinementAction> = {
         ],
         functions: [
           {
-            name: "write_in_language",
+            name: "log_translated_text",
             parameters: {
               type: "object",
               properties: {
-                text: {
+                translated_text: {
                   type: "string",
+                  description: "The text, written in the language specified in the prompt",
                 },
               },
             },
           },
         ],
         function_call: {
-          name: "write_in_language",
+          name: "log_translated_text",
         },
       });
+
+      =========
   
       Add an OpenAI function that takes one or more nested parameters that match the expected output from this prompt.`,
   },

app/src/pages/api/sentry-example-api.js

@@ -1,6 +0,0 @@
-// A faulty API route to test Sentry's error monitoring
-// @ts-expect-error just a test file, don't care about types
-export default function handler(_req, res) {
-  throw new Error("Sentry Example API Route Error");
-  res.status(200).json({ name: "John Doe" });
-}

app/src/pages/api/v1/[...trpc].ts

@@ -1,17 +1,14 @@
 import { type NextApiRequest, type NextApiResponse } from "next";
 import cors from "nextjs-cors";
 import { createOpenApiNextHandler } from "trpc-openapi";
-import { createProcedureCache } from "trpc-openapi/dist/adapters/node-http/procedures";
-import { appRouter } from "~/server/api/root.router";
-import { createTRPCContext } from "~/server/api/trpc";
+import { v1ApiRouter } from "~/server/api/external/v1Api.router";
+import { createOpenApiContext } from "~/server/api/external/openApiTrpc";
 
 const openApiHandler = createOpenApiNextHandler({
-  router: appRouter,
-  createContext: createTRPCContext,
+  router: v1ApiRouter,
+  createContext: createOpenApiContext,
 });
 
-const cache = createProcedureCache(appRouter);
-
 const handler = async (req: NextApiRequest, res: NextApiResponse) => {
   // Setup CORS
   await cors(req, res);

app/src/pages/api/v1/openapi.json.ts

@@ -1,12 +1,12 @@
 import { type NextApiRequest, type NextApiResponse } from "next";
 import { generateOpenApiDocument } from "trpc-openapi";
-import { appRouter } from "~/server/api/root.router";
+import { v1ApiRouter } from "~/server/api/external/v1Api.router";
 
-export const openApiDocument = generateOpenApiDocument(appRouter, {
+export const openApiDocument = generateOpenApiDocument(v1ApiRouter, {
   title: "OpenPipe API",
   description: "The public API for reporting API calls to OpenPipe",
-  version: "0.1.0",
-  baseUrl: "https://app.openpipe.ai/api",
+  version: "0.1.1",
+  baseUrl: "https://app.openpipe.ai/api/v1",
 });
 // Respond with our OpenAPI schema
 const hander = (req: NextApiRequest, res: NextApiResponse) => {

app/src/server/api/external/openApiTrpc.ts

@@ -0,0 +1,95 @@
+import type { ApiKey, Project } from "@prisma/client";
+import { TRPCError, initTRPC } from "@trpc/server";
+import { type CreateNextContextOptions } from "@trpc/server/adapters/next";
+import superjson from "superjson";
+import { type OpenApiMeta } from "trpc-openapi";
+import { ZodError } from "zod";
+import { prisma } from "~/server/db";
+
+type CreateContextOptions = {
+  key:
+    | (ApiKey & {
+        project: Project;
+      })
+    | null;
+};
+
+/**
+ * This helper generates the "internals" for a tRPC context. If you need to use it, you can export
+ * it from here.
+ *
+ * Examples of things you may need it for:
+ * - testing, so we don't have to mock Next.js' req/res
+ * - tRPC's `createSSGHelpers`, where we don't have req/res
+ *
+ * @see https://create.t3.gg/en/usage/trpc#-serverapitrpcts
+ */
+export const createInnerTRPCContext = (opts: CreateContextOptions) => {
+  return {
+    key: opts.key,
+  };
+};
+
+export const createOpenApiContext = async (opts: CreateNextContextOptions) => {
+  const { req, res } = opts;
+
+  const apiKey = req.headers.authorization?.split(" ")[1] as string | null;
+
+  if (!apiKey) {
+    throw new TRPCError({ code: "UNAUTHORIZED" });
+  }
+  const key = await prisma.apiKey.findUnique({
+    where: { apiKey },
+    include: { project: true },
+  });
+  if (!key) {
+    throw new TRPCError({ code: "UNAUTHORIZED" });
+  }
+
+  return createInnerTRPCContext({
+    key,
+  });
+};
+
+export type TRPCContext = Awaited<ReturnType<typeof createOpenApiContext>>;
+
+const t = initTRPC
+  .context<typeof createOpenApiContext>()
+  .meta<OpenApiMeta>()
+  .create({
+    transformer: superjson,
+    errorFormatter({ shape, error }) {
+      return {
+        ...shape,
+        data: {
+          ...shape.data,
+          zodError: error.cause instanceof ZodError ? error.cause.flatten() : null,
+        },
+      };
+    },
+  });
+
+export const createOpenApiRouter = t.router;
+
+export const openApiPublicProc = t.procedure;
+
+/** Reusable middleware that enforces users are logged in before running the procedure. */
+const enforceApiKey = t.middleware(async ({ ctx, next }) => {
+  if (!ctx.key) {
+    throw new TRPCError({ code: "UNAUTHORIZED" });
+  }
+
+  return next({
+    ctx: { key: ctx.key },
+  });
+});
+
+/**
+ * Protected (authenticated) procedure
+ *
+ * If you want a query or mutation to ONLY be accessible to logged in users, use this. It verifies
+ * the session is valid and guarantees `ctx.session.user` is not null.
+ *
+ * @see https://trpc.io/docs/procedures
+ */
+export const openApiProtectedProc = t.procedure.use(enforceApiKey);

app/src/server/api/external/v1Api.router.ts

@@ -2,9 +2,6 @@ import { type Prisma } from "@prisma/client";
 import { type JsonValue } from "type-fest";
 import { z } from "zod";
 import { v4 as uuidv4 } from "uuid";
-import { TRPCError } from "@trpc/server";
-
-import { createTRPCRouter, publicProcedure } from "~/server/api/trpc";
 import { prisma } from "~/server/db";
 import { hashRequest } from "~/server/utils/hashObject";
 import modelProvider from "~/modelProviders/openai-ChatCompletion";
@@ -12,6 +9,7 @@ import {
   type ChatCompletion,
   type CompletionCreateParams,
 } from "openai/resources/chat/completions";
+import { createOpenApiRouter, openApiProtectedProc } from "./openApiTrpc";
 
 const reqValidator = z.object({
   model: z.string(),
@@ -28,12 +26,12 @@ const respValidator = z.object({
   ),
 });
 
-export const externalApiRouter = createTRPCRouter({
-  checkCache: publicProcedure
+export const v1ApiRouter = createOpenApiRouter({
+  checkCache: openApiProtectedProc
     .meta({
       openapi: {
         method: "POST",
-        path: "/v1/check-cache",
+        path: "/check-cache",
         description: "Check if a prompt is cached",
         protect: true,
       },
@@ -47,7 +45,8 @@ export const externalApiRouter = createTRPCRouter({
           .optional()
           .describe(
             'Extra tags to attach to the call for filtering. Eg { "userId": "123", "promptId": "populate-title" }',
-          ),
+          )
+          .default({}),
       }),
     )
     .output(
@@ -56,18 +55,8 @@ export const externalApiRouter = createTRPCRouter({
       }),
     )
     .mutation(async ({ input, ctx }) => {
-      const apiKey = ctx.apiKey;
-      if (!apiKey) {
-        throw new TRPCError({ code: "UNAUTHORIZED" });
-      }
-      const key = await prisma.apiKey.findUnique({
-        where: { apiKey },
-      });
-      if (!key) {
-        throw new TRPCError({ code: "UNAUTHORIZED" });
-      }
       const reqPayload = await reqValidator.spa(input.reqPayload);
-      const cacheKey = hashRequest(key.projectId, reqPayload as JsonValue);
+      const cacheKey = hashRequest(ctx.key.projectId, reqPayload as JsonValue);
 
       const existingResponse = await prisma.loggedCallModelResponse.findFirst({
         where: { cacheKey },
@@ -79,23 +68,24 @@ export const externalApiRouter = createTRPCRouter({
 
       await prisma.loggedCall.create({
         data: {
-          projectId: key.projectId,
+          projectId: ctx.key.projectId,
           requestedAt: new Date(input.requestedAt),
           cacheHit: true,
           modelResponseId: existingResponse.id,
         },
       });
 
+      await createTags(existingResponse.originalLoggedCallId, input.tags);
       return {
         respPayload: existingResponse.respPayload,
       };
     }),
 
-  report: publicProcedure
+  report: openApiProtectedProc
     .meta({
       openapi: {
         method: "POST",
-        path: "/v1/report",
+        path: "/report",
         description: "Report an API call",
         protect: true,
       },
@@ -113,26 +103,16 @@ export const externalApiRouter = createTRPCRouter({
           .optional()
           .describe(
             'Extra tags to attach to the call for filtering. Eg { "userId": "123", "promptId": "populate-title" }',
-          ),
+          )
+          .default({}),
       }),
     )
     .output(z.void())
     .mutation(async ({ input, ctx }) => {
-      console.log("GOT TAGS", input.tags);
-      const apiKey = ctx.apiKey;
-      if (!apiKey) {
-        throw new TRPCError({ code: "UNAUTHORIZED" });
-      }
-      const key = await prisma.apiKey.findUnique({
-        where: { apiKey },
-      });
-      if (!key) {
-        throw new TRPCError({ code: "UNAUTHORIZED" });
-      }
       const reqPayload = await reqValidator.spa(input.reqPayload);
       const respPayload = await respValidator.spa(input.respPayload);
 
-      const requestHash = hashRequest(key.projectId, reqPayload as JsonValue);
+      const requestHash = hashRequest(ctx.key.projectId, reqPayload as JsonValue);
 
       const newLoggedCallId = uuidv4();
       const newModelResponseId = uuidv4();
@@ -151,7 +131,7 @@ export const externalApiRouter = createTRPCRouter({
         prisma.loggedCall.create({
           data: {
             id: newLoggedCallId,
-            projectId: key.projectId,
+            projectId: ctx.key.projectId,
             requestedAt: new Date(input.requestedAt),
             cacheHit: false,
             model,
@@ -185,14 +165,76 @@ export const externalApiRouter = createTRPCRouter({
         }),
       ]);
 
-      const tagsToCreate = Object.entries(input.tags ?? {}).map(([name, value]) => ({
-        loggedCallId: newLoggedCallId,
-        // sanitize tags
-        name: name.replaceAll(/[^a-zA-Z0-9_]/g, "_"),
-        value,
-      }));
-      await prisma.loggedCallTag.createMany({
-        data: tagsToCreate,
+      await createTags(newLoggedCallId, input.tags);
+    }),
+  localTestingOnlyGetLatestLoggedCall: openApiProtectedProc
+    .meta({
+      openapi: {
+        method: "GET",
+        path: "/local-testing-only-get-latest-logged-call",
+        description: "Get the latest logged call (only for local testing)",
+        protect: true, // Make sure to protect this endpoint
+      },
+    })
+    .input(z.void())
+    .output(
+      z
+        .object({
+          createdAt: z.date(),
+          cacheHit: z.boolean(),
+          tags: z.record(z.string().nullable()),
+          modelResponse: z
+            .object({
+              id: z.string(),
+              statusCode: z.number().nullable(),
+              errorMessage: z.string().nullable(),
+              reqPayload: z.unknown(),
+              respPayload: z.unknown(),
+            })
+            .nullable(),
+        })
+        .nullable(),
+    )
+    .mutation(async ({ ctx }) => {
+      if (process.env.NODE_ENV === "production") {
+        throw new Error("This operation is not allowed in production environment");
+      }
+
+      const latestLoggedCall = await prisma.loggedCall.findFirst({
+        where: { projectId: ctx.key.projectId },
+        orderBy: { requestedAt: "desc" },
+        select: {
+          createdAt: true,
+          cacheHit: true,
+          tags: true,
+          modelResponse: {
+            select: {
+              id: true,
+              statusCode: true,
+              errorMessage: true,
+              reqPayload: true,
+              respPayload: true,
+            },
+          },
+        },
       });
+
+      return (
+        latestLoggedCall && {
+          ...latestLoggedCall,
+          tags: Object.fromEntries(latestLoggedCall.tags.map((tag) => [tag.name, tag.value])),
+        }
+      );
     }),
 });
+
+async function createTags(loggedCallId: string, tags: Record<string, string>) {
+  const tagsToCreate = Object.entries(tags).map(([name, value]) => ({
+    loggedCallId,
+    name: name.replaceAll(/[^a-zA-Z0-9_$]/g, "_"),
+    value,
+  }));
+  await prisma.loggedCallTag.createMany({
+    data: tagsToCreate,
+  });
+}

app/src/server/api/root.router.ts

@@ -8,7 +8,6 @@ import { evaluationsRouter } from "./routers/evaluations.router";
 import { worldChampsRouter } from "./routers/worldChamps.router";
 import { datasetsRouter } from "./routers/datasets.router";
 import { datasetEntries } from "./routers/datasetEntries.router";
-import { externalApiRouter } from "./routers/externalApi.router";
 import { projectsRouter } from "./routers/projects.router";
 import { dashboardRouter } from "./routers/dashboard.router";
 import { loggedCallsRouter } from "./routers/loggedCalls.router";
@@ -31,7 +30,6 @@ export const appRouter = createTRPCRouter({
   projects: projectsRouter,
   dashboard: dashboardRouter,
   loggedCalls: loggedCallsRouter,
-  externalApi: externalApiRouter,
 });
 
 // export type definition of API

app/src/server/api/routers/promptVariants.router.ts

@@ -131,6 +131,8 @@ export const promptVariantsRouter = createTRPCRouter({
       const inputTokens = overallTokens._sum?.inputTokens ?? 0;
       const outputTokens = overallTokens._sum?.outputTokens ?? 0;
 
+      const awaitingCompletions = outputCount < scenarioCount;
+
       const awaitingEvals = !!evalResults.find(
         (result) => result.totalCount < scenarioCount * evals.length,
       );
@@ -142,6 +144,7 @@ export const promptVariantsRouter = createTRPCRouter({
         overallCost: overallTokens._sum?.cost ?? 0,
         scenarioCount,
         outputCount,
+        awaitingCompletions,
         awaitingEvals,
       };
     }),

app/src/server/api/trpc.ts

@@ -27,7 +27,6 @@ import { capturePath } from "~/utils/analytics/serverAnalytics";
 
 type CreateContextOptions = {
   session: Session | null;
-  apiKey: string | null;
 };
 
 // eslint-disable-next-line @typescript-eslint/no-empty-function
@@ -46,7 +45,6 @@ const noOp = () => {};
 export const createInnerTRPCContext = (opts: CreateContextOptions) => {
   return {
     session: opts.session,
-    apiKey: opts.apiKey,
     prisma,
     markAccessControlRun: noOp,
   };
@@ -64,11 +62,8 @@ export const createTRPCContext = async (opts: CreateNextContextOptions) => {
   // Get the session from the server using the getServerSession wrapper function
   const session = await getServerAuthSession({ req, res });
 
-  const apiKey = req.headers.authorization?.split(" ")[1] as string | null;
-
   return createInnerTRPCContext({
     session,
-    apiKey,
   });
 };
 

app/src/server/scripts/client-codegen.ts

@@ -1,5 +1,5 @@
 import "dotenv/config";
-import { openApiDocument } from "~/pages/api/openapi.json";
+import { openApiDocument } from "~/pages/api/v1/openapi.json";
 import fs from "fs";
 import path from "path";
 import { execSync } from "child_process";

app/src/server/utils/deriveNewContructFn.ts

@@ -51,7 +51,7 @@ const requestUpdatedPromptFunction = async (
             originalModelProvider.inputSchema,
             null,
             2,
-          )}\n\nDo not add any assistant messages.`,
+          )}`,
         },
         {
           role: "user",

app/src/server/utils/generateApiKey.ts

@@ -2,4 +2,4 @@ import cryptoRandomString from "crypto-random-string";
 
 const KEY_LENGTH = 42;
 
-export const generateApiKey = () => `opc_${cryptoRandomString({ length: KEY_LENGTH })}`;
+export const generateApiKey = () => `opk_${cryptoRandomString({ length: KEY_LENGTH })}`;

client-libs/openapi.json

@@ -3,17 +3,17 @@
   "info": {
     "title": "OpenPipe API",
     "description": "The public API for reporting API calls to OpenPipe",
-    "version": "0.1.0"
+    "version": "0.1.1"
   },
   "servers": [
     {
-      "url": "https://app.openpipe.ai/api"
+      "url": "https://app.openpipe.ai/api/v1"
     }
   ],
   "paths": {
-    "/v1/check-cache": {
+    "/check-cache": {
       "post": {
-        "operationId": "externalApi-checkCache",
+        "operationId": "checkCache",
         "description": "Check if a prompt is cached",
         "security": [
           {
@@ -39,7 +39,8 @@
                     "additionalProperties": {
                       "type": "string"
                     },
-                    "description": "Extra tags to attach to the call for filtering. Eg { \"userId\": \"123\", \"promptId\": \"populate-title\" }"
+                    "description": "Extra tags to attach to the call for filtering. Eg { \"userId\": \"123\", \"promptId\": \"populate-title\" }",
+                    "default": {}
                   }
                 },
                 "required": [
@@ -74,9 +75,9 @@
         }
       }
     },
-    "/v1/report": {
+    "/report": {
       "post": {
-        "operationId": "externalApi-report",
+        "operationId": "report",
         "description": "Report an API call",
         "security": [
           {
@@ -117,7 +118,8 @@
                     "additionalProperties": {
                       "type": "string"
                     },
-                    "description": "Extra tags to attach to the call for filtering. Eg { \"userId\": \"123\", \"promptId\": \"populate-title\" }"
+                    "description": "Extra tags to attach to the call for filtering. Eg { \"userId\": \"123\", \"promptId\": \"populate-title\" }",
+                    "default": {}
                   }
                 },
                 "required": [
@@ -144,6 +146,82 @@
           }
         }
       }
+    },
+    "/local-testing-only-get-latest-logged-call": {
+      "get": {
+        "operationId": "localTestingOnlyGetLatestLoggedCall",
+        "description": "Get the latest logged call (only for local testing)",
+        "security": [
+          {
+            "Authorization": []
+          }
+        ],
+        "parameters": [],
+        "responses": {
+          "200": {
+            "description": "Successful response",
+            "content": {
+              "application/json": {
+                "schema": {
+                  "type": "object",
+                  "properties": {
+                    "createdAt": {
+                      "type": "string",
+                      "format": "date-time"
+                    },
+                    "cacheHit": {
+                      "type": "boolean"
+                    },
+                    "tags": {
+                      "type": "object",
+                      "additionalProperties": {
+                        "type": "string",
+                        "nullable": true
+                      }
+                    },
+                    "modelResponse": {
+                      "type": "object",
+                      "properties": {
+                        "id": {
+                          "type": "string"
+                        },
+                        "statusCode": {
+                          "type": "number",
+                          "nullable": true
+                        },
+                        "errorMessage": {
+                          "type": "string",
+                          "nullable": true
+                        },
+                        "reqPayload": {},
+                        "respPayload": {}
+                      },
+                      "required": [
+                        "id",
+                        "statusCode",
+                        "errorMessage"
+                      ],
+                      "additionalProperties": false,
+                      "nullable": true
+                    }
+                  },
+                  "required": [
+                    "createdAt",
+                    "cacheHit",
+                    "tags",
+                    "modelResponse"
+                  ],
+                  "additionalProperties": false,
+                  "nullable": true
+                }
+              }
+            }
+          },
+          "default": {
+            "$ref": "#/components/responses/error"
+          }
+        }
+      }
     }
   },
   "components": {

client-libs/python/openpipe/api_client/api/default/check_cache.py

@@ -5,14 +5,14 @@ import httpx
 
 from ... import errors
 from ...client import AuthenticatedClient, Client
-from ...models.external_api_check_cache_json_body import ExternalApiCheckCacheJsonBody
-from ...models.external_api_check_cache_response_200 import ExternalApiCheckCacheResponse200
+from ...models.check_cache_json_body import CheckCacheJsonBody
+from ...models.check_cache_response_200 import CheckCacheResponse200
 from ...types import Response
 
 
 def _get_kwargs(
     *,
-    json_body: ExternalApiCheckCacheJsonBody,
+    json_body: CheckCacheJsonBody,
 ) -> Dict[str, Any]:
     pass
 
@@ -20,16 +20,16 @@ def _get_kwargs(
 
     return {
         "method": "post",
-        "url": "/v1/check-cache",
+        "url": "/check-cache",
         "json": json_json_body,
     }
 
 
 def _parse_response(
     *, client: Union[AuthenticatedClient, Client], response: httpx.Response
-) -> Optional[ExternalApiCheckCacheResponse200]:
+) -> Optional[CheckCacheResponse200]:
     if response.status_code == HTTPStatus.OK:
-        response_200 = ExternalApiCheckCacheResponse200.from_dict(response.json())
+        response_200 = CheckCacheResponse200.from_dict(response.json())
 
         return response_200
     if client.raise_on_unexpected_status:
@@ -40,7 +40,7 @@ def _parse_response(
 
 def _build_response(
     *, client: Union[AuthenticatedClient, Client], response: httpx.Response
-) -> Response[ExternalApiCheckCacheResponse200]:
+) -> Response[CheckCacheResponse200]:
     return Response(
         status_code=HTTPStatus(response.status_code),
         content=response.content,
@@ -52,19 +52,19 @@ def _build_response(
 def sync_detailed(
     *,
     client: AuthenticatedClient,
-    json_body: ExternalApiCheckCacheJsonBody,
-) -> Response[ExternalApiCheckCacheResponse200]:
+    json_body: CheckCacheJsonBody,
+) -> Response[CheckCacheResponse200]:
     """Check if a prompt is cached
 
     Args:
-        json_body (ExternalApiCheckCacheJsonBody):
+        json_body (CheckCacheJsonBody):
 
     Raises:
         errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
         httpx.TimeoutException: If the request takes longer than Client.timeout.
 
     Returns:
-        Response[ExternalApiCheckCacheResponse200]
+        Response[CheckCacheResponse200]
     """
 
     kwargs = _get_kwargs(
@@ -81,19 +81,19 @@ def sync_detailed(
 def sync(
     *,
     client: AuthenticatedClient,
-    json_body: ExternalApiCheckCacheJsonBody,
-) -> Optional[ExternalApiCheckCacheResponse200]:
+    json_body: CheckCacheJsonBody,
+) -> Optional[CheckCacheResponse200]:
     """Check if a prompt is cached
 
     Args:
-        json_body (ExternalApiCheckCacheJsonBody):
+        json_body (CheckCacheJsonBody):
 
     Raises:
         errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
         httpx.TimeoutException: If the request takes longer than Client.timeout.
 
     Returns:
-        ExternalApiCheckCacheResponse200
+        CheckCacheResponse200
     """
 
     return sync_detailed(
@@ -105,19 +105,19 @@ def sync(
 async def asyncio_detailed(
     *,
     client: AuthenticatedClient,
-    json_body: ExternalApiCheckCacheJsonBody,
-) -> Response[ExternalApiCheckCacheResponse200]:
+    json_body: CheckCacheJsonBody,
+) -> Response[CheckCacheResponse200]:
     """Check if a prompt is cached
 
     Args:
-        json_body (ExternalApiCheckCacheJsonBody):
+        json_body (CheckCacheJsonBody):
 
     Raises:
         errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
         httpx.TimeoutException: If the request takes longer than Client.timeout.
 
     Returns:
-        Response[ExternalApiCheckCacheResponse200]
+        Response[CheckCacheResponse200]
     """
 
     kwargs = _get_kwargs(
@@ -132,19 +132,19 @@ async def asyncio_detailed(
 async def asyncio(
     *,
     client: AuthenticatedClient,
-    json_body: ExternalApiCheckCacheJsonBody,
-) -> Optional[ExternalApiCheckCacheResponse200]:
+    json_body: CheckCacheJsonBody,
+) -> Optional[CheckCacheResponse200]:
     """Check if a prompt is cached
 
     Args:
-        json_body (ExternalApiCheckCacheJsonBody):
+        json_body (CheckCacheJsonBody):
 
     Raises:
         errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
         httpx.TimeoutException: If the request takes longer than Client.timeout.
 
     Returns:
-        ExternalApiCheckCacheResponse200
+        CheckCacheResponse200
     """
 
     return (

client-libs/python/openpipe/api_client/api/default/local_testing_only_get_latest_logged_call.py

@@ -0,0 +1,133 @@
+from http import HTTPStatus
+from typing import Any, Dict, Optional, Union
+
+import httpx
+
+from ... import errors
+from ...client import AuthenticatedClient, Client
+from ...models.local_testing_only_get_latest_logged_call_response_200 import (
+    LocalTestingOnlyGetLatestLoggedCallResponse200,
+)
+from ...types import Response
+
+
+def _get_kwargs() -> Dict[str, Any]:
+    pass
+
+    return {
+        "method": "get",
+        "url": "/local-testing-only-get-latest-logged-call",
+    }
+
+
+def _parse_response(
+    *, client: Union[AuthenticatedClient, Client], response: httpx.Response
+) -> Optional[Optional[LocalTestingOnlyGetLatestLoggedCallResponse200]]:
+    if response.status_code == HTTPStatus.OK:
+        _response_200 = response.json()
+        response_200: Optional[LocalTestingOnlyGetLatestLoggedCallResponse200]
+        if _response_200 is None:
+            response_200 = None
+        else:
+            response_200 = LocalTestingOnlyGetLatestLoggedCallResponse200.from_dict(_response_200)
+
+        return response_200
+    if client.raise_on_unexpected_status:
+        raise errors.UnexpectedStatus(response.status_code, response.content)
+    else:
+        return None
+
+
+def _build_response(
+    *, client: Union[AuthenticatedClient, Client], response: httpx.Response
+) -> Response[Optional[LocalTestingOnlyGetLatestLoggedCallResponse200]]:
+    return Response(
+        status_code=HTTPStatus(response.status_code),
+        content=response.content,
+        headers=response.headers,
+        parsed=_parse_response(client=client, response=response),
+    )
+
+
+def sync_detailed(
+    *,
+    client: AuthenticatedClient,
+) -> Response[Optional[LocalTestingOnlyGetLatestLoggedCallResponse200]]:
+    """Get the latest logged call (only for local testing)
+
+    Raises:
+        errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
+        httpx.TimeoutException: If the request takes longer than Client.timeout.
+
+    Returns:
+        Response[Optional[LocalTestingOnlyGetLatestLoggedCallResponse200]]
+    """
+
+    kwargs = _get_kwargs()
+
+    response = client.get_httpx_client().request(
+        **kwargs,
+    )
+
+    return _build_response(client=client, response=response)
+
+
+def sync(
+    *,
+    client: AuthenticatedClient,
+) -> Optional[Optional[LocalTestingOnlyGetLatestLoggedCallResponse200]]:
+    """Get the latest logged call (only for local testing)
+
+    Raises:
+        errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
+        httpx.TimeoutException: If the request takes longer than Client.timeout.
+
+    Returns:
+        Optional[LocalTestingOnlyGetLatestLoggedCallResponse200]
+    """
+
+    return sync_detailed(
+        client=client,
+    ).parsed
+
+
+async def asyncio_detailed(
+    *,
+    client: AuthenticatedClient,
+) -> Response[Optional[LocalTestingOnlyGetLatestLoggedCallResponse200]]:
+    """Get the latest logged call (only for local testing)
+
+    Raises:
+        errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
+        httpx.TimeoutException: If the request takes longer than Client.timeout.
+
+    Returns:
+        Response[Optional[LocalTestingOnlyGetLatestLoggedCallResponse200]]
+    """
+
+    kwargs = _get_kwargs()
+
+    response = await client.get_async_httpx_client().request(**kwargs)
+
+    return _build_response(client=client, response=response)
+
+
+async def asyncio(
+    *,
+    client: AuthenticatedClient,
+) -> Optional[Optional[LocalTestingOnlyGetLatestLoggedCallResponse200]]:
+    """Get the latest logged call (only for local testing)
+
+    Raises:
+        errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
+        httpx.TimeoutException: If the request takes longer than Client.timeout.
+
+    Returns:
+        Optional[LocalTestingOnlyGetLatestLoggedCallResponse200]
+    """
+
+    return (
+        await asyncio_detailed(
+            client=client,
+        )
+    ).parsed

client-libs/python/openpipe/api_client/api/default/report.py

@@ -5,13 +5,13 @@ import httpx
 
 from ... import errors
 from ...client import AuthenticatedClient, Client
-from ...models.external_api_report_json_body import ExternalApiReportJsonBody
+from ...models.report_json_body import ReportJsonBody
 from ...types import Response
 
 
 def _get_kwargs(
     *,
-    json_body: ExternalApiReportJsonBody,
+    json_body: ReportJsonBody,
 ) -> Dict[str, Any]:
     pass
 
@@ -19,7 +19,7 @@ def _get_kwargs(
 
     return {
         "method": "post",
-        "url": "/v1/report",
+        "url": "/report",
         "json": json_json_body,
     }
 
@@ -45,12 +45,12 @@ def _build_response(*, client: Union[AuthenticatedClient, Client], response: htt
 def sync_detailed(
     *,
     client: AuthenticatedClient,
-    json_body: ExternalApiReportJsonBody,
+    json_body: ReportJsonBody,
 ) -> Response[Any]:
     """Report an API call
 
     Args:
-        json_body (ExternalApiReportJsonBody):
+        json_body (ReportJsonBody):
 
     Raises:
         errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
@@ -74,12 +74,12 @@ def sync_detailed(
 async def asyncio_detailed(
     *,
     client: AuthenticatedClient,
-    json_body: ExternalApiReportJsonBody,
+    json_body: ReportJsonBody,
 ) -> Response[Any]:
     """Report an API call
 
     Args:
-        json_body (ExternalApiReportJsonBody):
+        json_body (ReportJsonBody):
 
     Raises:
         errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.

client-libs/python/openpipe/api_client/models/__init__.py

@@ -1,15 +1,25 @@
 """ Contains all the data models used in inputs/outputs """
 
-from .external_api_check_cache_json_body import ExternalApiCheckCacheJsonBody
-from .external_api_check_cache_json_body_tags import ExternalApiCheckCacheJsonBodyTags
-from .external_api_check_cache_response_200 import ExternalApiCheckCacheResponse200
-from .external_api_report_json_body import ExternalApiReportJsonBody
-from .external_api_report_json_body_tags import ExternalApiReportJsonBodyTags
+from .check_cache_json_body import CheckCacheJsonBody
+from .check_cache_json_body_tags import CheckCacheJsonBodyTags
+from .check_cache_response_200 import CheckCacheResponse200
+from .local_testing_only_get_latest_logged_call_response_200 import LocalTestingOnlyGetLatestLoggedCallResponse200
+from .local_testing_only_get_latest_logged_call_response_200_model_response import (
+    LocalTestingOnlyGetLatestLoggedCallResponse200ModelResponse,
+)
+from .local_testing_only_get_latest_logged_call_response_200_tags import (
+    LocalTestingOnlyGetLatestLoggedCallResponse200Tags,
+)
+from .report_json_body import ReportJsonBody
+from .report_json_body_tags import ReportJsonBodyTags
 
 __all__ = (
-    "ExternalApiCheckCacheJsonBody",
-    "ExternalApiCheckCacheJsonBodyTags",
-    "ExternalApiCheckCacheResponse200",
-    "ExternalApiReportJsonBody",
-    "ExternalApiReportJsonBodyTags",
+    "CheckCacheJsonBody",
+    "CheckCacheJsonBodyTags",
+    "CheckCacheResponse200",
+    "LocalTestingOnlyGetLatestLoggedCallResponse200",
+    "LocalTestingOnlyGetLatestLoggedCallResponse200ModelResponse",
+    "LocalTestingOnlyGetLatestLoggedCallResponse200Tags",
+    "ReportJsonBody",
+    "ReportJsonBodyTags",
 )

client-libs/python/openpipe/api_client/models/check_cache_json_body.py

@@ -5,25 +5,25 @@ from attrs import define
 from ..types import UNSET, Unset
 
 if TYPE_CHECKING:
-    from ..models.external_api_check_cache_json_body_tags import ExternalApiCheckCacheJsonBodyTags
+    from ..models.check_cache_json_body_tags import CheckCacheJsonBodyTags
 
 
-T = TypeVar("T", bound="ExternalApiCheckCacheJsonBody")
+T = TypeVar("T", bound="CheckCacheJsonBody")
 
 
 @define
-class ExternalApiCheckCacheJsonBody:
+class CheckCacheJsonBody:
     """
     Attributes:
         requested_at (float): Unix timestamp in milliseconds
         req_payload (Union[Unset, Any]): JSON-encoded request payload
-        tags (Union[Unset, ExternalApiCheckCacheJsonBodyTags]): Extra tags to attach to the call for filtering. Eg {
-            "userId": "123", "promptId": "populate-title" }
+        tags (Union[Unset, CheckCacheJsonBodyTags]): Extra tags to attach to the call for filtering. Eg { "userId":
+            "123", "promptId": "populate-title" }
     """
 
     requested_at: float
     req_payload: Union[Unset, Any] = UNSET
-    tags: Union[Unset, "ExternalApiCheckCacheJsonBodyTags"] = UNSET
+    tags: Union[Unset, "CheckCacheJsonBodyTags"] = UNSET
 
     def to_dict(self) -> Dict[str, Any]:
         requested_at = self.requested_at
@@ -47,7 +47,7 @@ class ExternalApiCheckCacheJsonBody:
 
     @classmethod
     def from_dict(cls: Type[T], src_dict: Dict[str, Any]) -> T:
-        from ..models.external_api_check_cache_json_body_tags import ExternalApiCheckCacheJsonBodyTags
+        from ..models.check_cache_json_body_tags import CheckCacheJsonBodyTags
 
         d = src_dict.copy()
         requested_at = d.pop("requestedAt")
@@ -55,16 +55,16 @@ class ExternalApiCheckCacheJsonBody:
         req_payload = d.pop("reqPayload", UNSET)
 
         _tags = d.pop("tags", UNSET)
-        tags: Union[Unset, ExternalApiCheckCacheJsonBodyTags]
+        tags: Union[Unset, CheckCacheJsonBodyTags]
         if isinstance(_tags, Unset):
             tags = UNSET
         else:
-            tags = ExternalApiCheckCacheJsonBodyTags.from_dict(_tags)
+            tags = CheckCacheJsonBodyTags.from_dict(_tags)
 
-        external_api_check_cache_json_body = cls(
+        check_cache_json_body = cls(
             requested_at=requested_at,
             req_payload=req_payload,
             tags=tags,
         )
 
-        return external_api_check_cache_json_body
+        return check_cache_json_body

client-libs/python/openpipe/api_client/models/check_cache_json_body_tags.py

@@ -2,11 +2,11 @@ from typing import Any, Dict, List, Type, TypeVar
 
 from attrs import define, field
 
-T = TypeVar("T", bound="ExternalApiReportJsonBodyTags")
+T = TypeVar("T", bound="CheckCacheJsonBodyTags")
 
 
 @define
-class ExternalApiReportJsonBodyTags:
+class CheckCacheJsonBodyTags:
     """Extra tags to attach to the call for filtering. Eg { "userId": "123", "promptId": "populate-title" }"""
 
     additional_properties: Dict[str, str] = field(init=False, factory=dict)
@@ -21,10 +21,10 @@ class ExternalApiReportJsonBodyTags:
     @classmethod
     def from_dict(cls: Type[T], src_dict: Dict[str, Any]) -> T:
         d = src_dict.copy()
-        external_api_report_json_body_tags = cls()
+        check_cache_json_body_tags = cls()
 
-        external_api_report_json_body_tags.additional_properties = d
-        return external_api_report_json_body_tags
+        check_cache_json_body_tags.additional_properties = d
+        return check_cache_json_body_tags
 
     @property
     def additional_keys(self) -> List[str]:

client-libs/python/openpipe/api_client/models/check_cache_response_200.py

@@ -4,11 +4,11 @@ from attrs import define
 
 from ..types import UNSET, Unset
 
-T = TypeVar("T", bound="ExternalApiCheckCacheResponse200")
+T = TypeVar("T", bound="CheckCacheResponse200")
 
 
 @define
-class ExternalApiCheckCacheResponse200:
+class CheckCacheResponse200:
     """
     Attributes:
         resp_payload (Union[Unset, Any]): JSON-encoded response payload
@@ -31,8 +31,8 @@ class ExternalApiCheckCacheResponse200:
         d = src_dict.copy()
         resp_payload = d.pop("respPayload", UNSET)
 
-        external_api_check_cache_response_200 = cls(
+        check_cache_response_200 = cls(
             resp_payload=resp_payload,
         )
 
-        return external_api_check_cache_response_200
+        return check_cache_response_200

client-libs/python/openpipe/api_client/models/local_testing_only_get_latest_logged_call_response_200.py

@@ -0,0 +1,84 @@
+import datetime
+from typing import TYPE_CHECKING, Any, Dict, Optional, Type, TypeVar
+
+from attrs import define
+from dateutil.parser import isoparse
+
+if TYPE_CHECKING:
+    from ..models.local_testing_only_get_latest_logged_call_response_200_model_response import (
+        LocalTestingOnlyGetLatestLoggedCallResponse200ModelResponse,
+    )
+    from ..models.local_testing_only_get_latest_logged_call_response_200_tags import (
+        LocalTestingOnlyGetLatestLoggedCallResponse200Tags,
+    )
+
+
+T = TypeVar("T", bound="LocalTestingOnlyGetLatestLoggedCallResponse200")
+
+
+@define
+class LocalTestingOnlyGetLatestLoggedCallResponse200:
+    """
+    Attributes:
+        created_at (datetime.datetime):
+        cache_hit (bool):
+        tags (LocalTestingOnlyGetLatestLoggedCallResponse200Tags):
+        model_response (Optional[LocalTestingOnlyGetLatestLoggedCallResponse200ModelResponse]):
+    """
+
+    created_at: datetime.datetime
+    cache_hit: bool
+    tags: "LocalTestingOnlyGetLatestLoggedCallResponse200Tags"
+    model_response: Optional["LocalTestingOnlyGetLatestLoggedCallResponse200ModelResponse"]
+
+    def to_dict(self) -> Dict[str, Any]:
+        created_at = self.created_at.isoformat()
+
+        cache_hit = self.cache_hit
+        tags = self.tags.to_dict()
+
+        model_response = self.model_response.to_dict() if self.model_response else None
+
+        field_dict: Dict[str, Any] = {}
+        field_dict.update(
+            {
+                "createdAt": created_at,
+                "cacheHit": cache_hit,
+                "tags": tags,
+                "modelResponse": model_response,
+            }
+        )
+
+        return field_dict
+
+    @classmethod
+    def from_dict(cls: Type[T], src_dict: Dict[str, Any]) -> T:
+        from ..models.local_testing_only_get_latest_logged_call_response_200_model_response import (
+            LocalTestingOnlyGetLatestLoggedCallResponse200ModelResponse,
+        )
+        from ..models.local_testing_only_get_latest_logged_call_response_200_tags import (
+            LocalTestingOnlyGetLatestLoggedCallResponse200Tags,
+        )
+
+        d = src_dict.copy()
+        created_at = isoparse(d.pop("createdAt"))
+
+        cache_hit = d.pop("cacheHit")
+
+        tags = LocalTestingOnlyGetLatestLoggedCallResponse200Tags.from_dict(d.pop("tags"))
+
+        _model_response = d.pop("modelResponse")
+        model_response: Optional[LocalTestingOnlyGetLatestLoggedCallResponse200ModelResponse]
+        if _model_response is None:
+            model_response = None
+        else:
+            model_response = LocalTestingOnlyGetLatestLoggedCallResponse200ModelResponse.from_dict(_model_response)
+
+        local_testing_only_get_latest_logged_call_response_200 = cls(
+            created_at=created_at,
+            cache_hit=cache_hit,
+            tags=tags,
+            model_response=model_response,
+        )
+
+        return local_testing_only_get_latest_logged_call_response_200

client-libs/python/openpipe/api_client/models/local_testing_only_get_latest_logged_call_response_200_model_response.py

@@ -0,0 +1,70 @@
+from typing import Any, Dict, Optional, Type, TypeVar, Union
+
+from attrs import define
+
+from ..types import UNSET, Unset
+
+T = TypeVar("T", bound="LocalTestingOnlyGetLatestLoggedCallResponse200ModelResponse")
+
+
+@define
+class LocalTestingOnlyGetLatestLoggedCallResponse200ModelResponse:
+    """
+    Attributes:
+        id (str):
+        status_code (Optional[float]):
+        error_message (Optional[str]):
+        req_payload (Union[Unset, Any]):
+        resp_payload (Union[Unset, Any]):
+    """
+
+    id: str
+    status_code: Optional[float]
+    error_message: Optional[str]
+    req_payload: Union[Unset, Any] = UNSET
+    resp_payload: Union[Unset, Any] = UNSET
+
+    def to_dict(self) -> Dict[str, Any]:
+        id = self.id
+        status_code = self.status_code
+        error_message = self.error_message
+        req_payload = self.req_payload
+        resp_payload = self.resp_payload
+
+        field_dict: Dict[str, Any] = {}
+        field_dict.update(
+            {
+                "id": id,
+                "statusCode": status_code,
+                "errorMessage": error_message,
+            }
+        )
+        if req_payload is not UNSET:
+            field_dict["reqPayload"] = req_payload
+        if resp_payload is not UNSET:
+            field_dict["respPayload"] = resp_payload
+
+        return field_dict
+
+    @classmethod
+    def from_dict(cls: Type[T], src_dict: Dict[str, Any]) -> T:
+        d = src_dict.copy()
+        id = d.pop("id")
+
+        status_code = d.pop("statusCode")
+
+        error_message = d.pop("errorMessage")
+
+        req_payload = d.pop("reqPayload", UNSET)
+
+        resp_payload = d.pop("respPayload", UNSET)
+
+        local_testing_only_get_latest_logged_call_response_200_model_response = cls(
+            id=id,
+            status_code=status_code,
+            error_message=error_message,
+            req_payload=req_payload,
+            resp_payload=resp_payload,
+        )
+
+        return local_testing_only_get_latest_logged_call_response_200_model_response

client-libs/python/openpipe/api_client/models/local_testing_only_get_latest_logged_call_response_200_tags.py

@@ -0,0 +1,43 @@
+from typing import Any, Dict, List, Optional, Type, TypeVar
+
+from attrs import define, field
+
+T = TypeVar("T", bound="LocalTestingOnlyGetLatestLoggedCallResponse200Tags")
+
+
+@define
+class LocalTestingOnlyGetLatestLoggedCallResponse200Tags:
+    """ """
+
+    additional_properties: Dict[str, Optional[str]] = field(init=False, factory=dict)
+
+    def to_dict(self) -> Dict[str, Any]:
+        field_dict: Dict[str, Any] = {}
+        field_dict.update(self.additional_properties)
+        field_dict.update({})
+
+        return field_dict
+
+    @classmethod
+    def from_dict(cls: Type[T], src_dict: Dict[str, Any]) -> T:
+        d = src_dict.copy()
+        local_testing_only_get_latest_logged_call_response_200_tags = cls()
+
+        local_testing_only_get_latest_logged_call_response_200_tags.additional_properties = d
+        return local_testing_only_get_latest_logged_call_response_200_tags
+
+    @property
+    def additional_keys(self) -> List[str]:
+        return list(self.additional_properties.keys())
+
+    def __getitem__(self, key: str) -> Optional[str]:
+        return self.additional_properties[key]
+
+    def __setitem__(self, key: str, value: Optional[str]) -> None:
+        self.additional_properties[key] = value
+
+    def __delitem__(self, key: str) -> None:
+        del self.additional_properties[key]
+
+    def __contains__(self, key: str) -> bool:
+        return key in self.additional_properties

client-libs/python/openpipe/api_client/models/report_json_body.py

@@ -5,14 +5,14 @@ from attrs import define
 from ..types import UNSET, Unset
 
 if TYPE_CHECKING:
-    from ..models.external_api_report_json_body_tags import ExternalApiReportJsonBodyTags
+    from ..models.report_json_body_tags import ReportJsonBodyTags
 
 
-T = TypeVar("T", bound="ExternalApiReportJsonBody")
+T = TypeVar("T", bound="ReportJsonBody")
 
 
 @define
-class ExternalApiReportJsonBody:
+class ReportJsonBody:
     """
     Attributes:
         requested_at (float): Unix timestamp in milliseconds
@@ -21,8 +21,8 @@ class ExternalApiReportJsonBody:
         resp_payload (Union[Unset, Any]): JSON-encoded response payload
         status_code (Union[Unset, float]): HTTP status code of response
         error_message (Union[Unset, str]): User-friendly error message
-        tags (Union[Unset, ExternalApiReportJsonBodyTags]): Extra tags to attach to the call for filtering. Eg {
-            "userId": "123", "promptId": "populate-title" }
+        tags (Union[Unset, ReportJsonBodyTags]): Extra tags to attach to the call for filtering. Eg { "userId": "123",
+            "promptId": "populate-title" }
     """
 
     requested_at: float
@@ -31,7 +31,7 @@ class ExternalApiReportJsonBody:
     resp_payload: Union[Unset, Any] = UNSET
     status_code: Union[Unset, float] = UNSET
     error_message: Union[Unset, str] = UNSET
-    tags: Union[Unset, "ExternalApiReportJsonBodyTags"] = UNSET
+    tags: Union[Unset, "ReportJsonBodyTags"] = UNSET
 
     def to_dict(self) -> Dict[str, Any]:
         requested_at = self.requested_at
@@ -66,7 +66,7 @@ class ExternalApiReportJsonBody:
 
     @classmethod
     def from_dict(cls: Type[T], src_dict: Dict[str, Any]) -> T:
-        from ..models.external_api_report_json_body_tags import ExternalApiReportJsonBodyTags
+        from ..models.report_json_body_tags import ReportJsonBodyTags
 
         d = src_dict.copy()
         requested_at = d.pop("requestedAt")
@@ -82,13 +82,13 @@ class ExternalApiReportJsonBody:
         error_message = d.pop("errorMessage", UNSET)
 
         _tags = d.pop("tags", UNSET)
-        tags: Union[Unset, ExternalApiReportJsonBodyTags]
+        tags: Union[Unset, ReportJsonBodyTags]
         if isinstance(_tags, Unset):
             tags = UNSET
         else:
-            tags = ExternalApiReportJsonBodyTags.from_dict(_tags)
+            tags = ReportJsonBodyTags.from_dict(_tags)
 
-        external_api_report_json_body = cls(
+        report_json_body = cls(
             requested_at=requested_at,
             received_at=received_at,
             req_payload=req_payload,
@@ -98,4 +98,4 @@ class ExternalApiReportJsonBody:
             tags=tags,
         )
 
-        return external_api_report_json_body
+        return report_json_body

client-libs/python/openpipe/api_client/models/report_json_body_tags.py

@@ -2,11 +2,11 @@ from typing import Any, Dict, List, Type, TypeVar
 
 from attrs import define, field
 
-T = TypeVar("T", bound="ExternalApiCheckCacheJsonBodyTags")
+T = TypeVar("T", bound="ReportJsonBodyTags")
 
 
 @define
-class ExternalApiCheckCacheJsonBodyTags:
+class ReportJsonBodyTags:
     """Extra tags to attach to the call for filtering. Eg { "userId": "123", "promptId": "populate-title" }"""
 
     additional_properties: Dict[str, str] = field(init=False, factory=dict)
@@ -21,10 +21,10 @@ class ExternalApiCheckCacheJsonBodyTags:
     @classmethod
     def from_dict(cls: Type[T], src_dict: Dict[str, Any]) -> T:
         d = src_dict.copy()
-        external_api_check_cache_json_body_tags = cls()
+        report_json_body_tags = cls()
 
-        external_api_check_cache_json_body_tags.additional_properties = d
-        return external_api_check_cache_json_body_tags
+        report_json_body_tags.additional_properties = d
+        return report_json_body_tags
 
     @property
     def additional_keys(self) -> List[str]:

client-libs/python/openpipe/merge_openai_chunks.py

@@ -1,9 +1,9 @@
 from typing import Any, Optional
 
 
-def merge_streamed_chunks(base: Optional[Any], chunk: Any) -> Any:
+def merge_openai_chunks(base: Optional[Any], chunk: Any) -> Any:
     if base is None:
-        return merge_streamed_chunks({**chunk, "choices": []}, chunk)
+        return merge_openai_chunks({**chunk, "choices": []}, chunk)
 
     choices = base["choices"].copy()
     for choice in chunk["choices"]:
@@ -34,9 +34,7 @@ def merge_streamed_chunks(base: Optional[Any], chunk: Any) -> Any:
                 {**new_choice, "message": {"role": "assistant", **choice["delta"]}}
             )
 
-    merged = {
+    return {
         **base,
         "choices": choices,
     }
-
-    return merged

client-libs/python/openpipe/openai.py

@@ -3,9 +3,16 @@ from openai.openai_object import OpenAIObject
 import time
 import inspect
 
-from openpipe.merge_openai_chunks import merge_streamed_chunks
+from openpipe.merge_openai_chunks import merge_openai_chunks
+from openpipe.openpipe_meta import OpenPipeMeta
 
-from .shared import maybe_check_cache, maybe_check_cache_async, report_async, report
+from .shared import (
+    _should_check_cache,
+    maybe_check_cache,
+    maybe_check_cache_async,
+    report_async,
+    report,
+)
 
 
 class WrappedChatCompletion(original_openai.ChatCompletion):
@@ -29,9 +36,15 @@ class WrappedChatCompletion(original_openai.ChatCompletion):
                 def _gen():
                     assembled_completion = None
                     for chunk in chat_completion:
-                        assembled_completion = merge_streamed_chunks(
+                        assembled_completion = merge_openai_chunks(
                             assembled_completion, chunk
                         )
+
+                        cache_status = (
+                            "MISS" if _should_check_cache(openpipe_options) else "SKIP"
+                        )
+                        chunk.openpipe = OpenPipeMeta(cache_status=cache_status)
+
                         yield chunk
 
                     received_at = int(time.time() * 1000)
@@ -58,6 +71,10 @@ class WrappedChatCompletion(original_openai.ChatCompletion):
                     status_code=200,
                 )
 
+                cache_status = (
+                    "MISS" if _should_check_cache(openpipe_options) else "SKIP"
+                )
+                chat_completion["openpipe"] = OpenPipeMeta(cache_status=cache_status)
             return chat_completion
         except Exception as e:
             received_at = int(time.time() * 1000)
@@ -96,21 +113,28 @@ class WrappedChatCompletion(original_openai.ChatCompletion):
         requested_at = int(time.time() * 1000)
 
         try:
-            chat_completion = original_openai.ChatCompletion.acreate(*args, **kwargs)
+            chat_completion = await original_openai.ChatCompletion.acreate(
+                *args, **kwargs
+            )
 
-            if inspect.isgenerator(chat_completion):
+            if inspect.isasyncgen(chat_completion):
 
-                def _gen():
+                async def _gen():
                     assembled_completion = None
-                    for chunk in chat_completion:
-                        assembled_completion = merge_streamed_chunks(
+                    async for chunk in chat_completion:
+                        assembled_completion = merge_openai_chunks(
                             assembled_completion, chunk
                         )
+                        cache_status = (
+                            "MISS" if _should_check_cache(openpipe_options) else "SKIP"
+                        )
+                        chunk.openpipe = OpenPipeMeta(cache_status=cache_status)
+
                         yield chunk
 
                     received_at = int(time.time() * 1000)
 
-                    report_async(
+                    await report_async(
                         openpipe_options=openpipe_options,
                         requested_at=requested_at,
                         received_at=received_at,
@@ -123,7 +147,7 @@ class WrappedChatCompletion(original_openai.ChatCompletion):
             else:
                 received_at = int(time.time() * 1000)
 
-                report_async(
+                await report_async(
                     openpipe_options=openpipe_options,
                     requested_at=requested_at,
                     received_at=received_at,
@@ -132,12 +156,17 @@ class WrappedChatCompletion(original_openai.ChatCompletion):
                     status_code=200,
                 )
 
+                cache_status = (
+                    "MISS" if _should_check_cache(openpipe_options) else "SKIP"
+                )
+                chat_completion["openpipe"] = OpenPipeMeta(cache_status=cache_status)
+
             return chat_completion
         except Exception as e:
             received_at = int(time.time() * 1000)
 
             if isinstance(e, original_openai.OpenAIError):
-                report_async(
+                await report_async(
                     openpipe_options=openpipe_options,
                     requested_at=requested_at,
                     received_at=received_at,
@@ -147,7 +176,7 @@ class WrappedChatCompletion(original_openai.ChatCompletion):
                     status_code=e.http_status,
                 )
             else:
-                report_async(
+                await report_async(
                     openpipe_options=openpipe_options,
                     requested_at=requested_at,
                     received_at=received_at,

client-libs/python/openpipe/openpipe_meta.py

@@ -0,0 +1,7 @@
+from attr import dataclass
+
+
+@dataclass
+class OpenPipeMeta:
+    # Cache status. One of 'HIT', 'MISS', 'SKIP'
+    cache_status: str

client-libs/python/openpipe/shared.py

@@ -1,10 +1,10 @@
 from openpipe.api_client.api.default import (
-    external_api_report,
-    external_api_check_cache,
+    report as api_report,
+    check_cache,
 )
 from openpipe.api_client.client import AuthenticatedClient
-from openpipe.api_client.models.external_api_report_json_body_tags import (
-    ExternalApiReportJsonBodyTags,
+from openpipe.api_client.models.report_json_body_tags import (
+    ReportJsonBodyTags,
 )
 import toml
 import time
@@ -21,7 +21,7 @@ def _get_tags(openpipe_options):
     tags["$sdk"] = "python"
     tags["$sdk_version"] = version
 
-    return ExternalApiReportJsonBodyTags.from_dict(tags)
+    return ReportJsonBodyTags.from_dict(tags)
 
 
 def _should_check_cache(openpipe_options):
@@ -31,7 +31,7 @@ def _should_check_cache(openpipe_options):
 
 
 def _process_cache_payload(
-    payload: external_api_check_cache.ExternalApiCheckCacheResponse200,
+    payload: check_cache.CheckCacheResponse200,
 ):
     if not payload or not payload.resp_payload:
         return None
@@ -47,9 +47,9 @@ def maybe_check_cache(
     if not _should_check_cache(openpipe_options):
         return None
     try:
-        payload = external_api_check_cache.sync(
+        payload = check_cache.sync(
             client=configured_client,
-            json_body=external_api_check_cache.ExternalApiCheckCacheJsonBody(
+            json_body=check_cache.CheckCacheJsonBody(
                 req_payload=req_payload,
                 requested_at=int(time.time() * 1000),
                 tags=_get_tags(openpipe_options),
@@ -72,9 +72,9 @@ async def maybe_check_cache_async(
         return None
 
     try:
-        payload = await external_api_check_cache.asyncio(
+        payload = await check_cache.asyncio(
             client=configured_client,
-            json_body=external_api_check_cache.ExternalApiCheckCacheJsonBody(
+            json_body=check_cache.CheckCacheJsonBody(
                 req_payload=req_payload,
                 requested_at=int(time.time() * 1000),
                 tags=_get_tags(openpipe_options),
@@ -94,9 +94,9 @@ def report(
     **kwargs,
 ):
     try:
-        external_api_report.sync_detailed(
+        api_report.sync_detailed(
             client=configured_client,
-            json_body=external_api_report.ExternalApiReportJsonBody(
+            json_body=api_report.ReportJsonBody(
                 **kwargs,
                 tags=_get_tags(openpipe_options),
             ),
@@ -112,9 +112,9 @@ async def report_async(
     **kwargs,
 ):
     try:
-        await external_api_report.asyncio_detailed(
+        await api_report.asyncio_detailed(
             client=configured_client,
-            json_body=external_api_report.ExternalApiReportJsonBody(
+            json_body=api_report.ReportJsonBody(
                 **kwargs,
                 tags=_get_tags(openpipe_options),
             ),

client-libs/python/openpipe/test_client.py

@@ -1,55 +1,106 @@
+from functools import reduce
 from dotenv import load_dotenv
-from . import openai, configure_openpipe
 import os
 import pytest
+from . import openai, configure_openpipe, configured_client
+from .api_client.api.default import local_testing_only_get_latest_logged_call
+from .merge_openai_chunks import merge_openai_chunks
+import random
+import string
+
+
+def random_string(length):
+    letters = string.ascii_lowercase
+    return "".join(random.choice(letters) for i in range(length))
+
 
 load_dotenv()
 
 openai.api_key = os.getenv("OPENAI_API_KEY")
 
 configure_openpipe(
-    base_url="http://localhost:3000/api", api_key=os.getenv("OPENPIPE_API_KEY")
+    base_url="http://localhost:3000/api/v1", api_key=os.getenv("OPENPIPE_API_KEY")
 )
 
 
+def last_logged_call():
+    return local_testing_only_get_latest_logged_call.sync(client=configured_client)
+
+
 def test_sync():
     completion = openai.ChatCompletion.create(
         model="gpt-3.5-turbo",
-        messages=[{"role": "system", "content": "count to 10"}],
+        messages=[{"role": "system", "content": "count to 3"}],
     )
 
-    print(completion.choices[0].message.content)
+    last_logged = last_logged_call()
+    assert (
+        last_logged.model_response.resp_payload["choices"][0]["message"]["content"]
+        == completion.choices[0].message.content
+    )
+    assert (
+        last_logged.model_response.req_payload["messages"][0]["content"] == "count to 3"
+    )
+
+    assert completion.openpipe.cache_status == "SKIP"
 
 
 def test_streaming():
     completion = openai.ChatCompletion.create(
         model="gpt-3.5-turbo",
-        messages=[{"role": "system", "content": "count to 10"}],
+        messages=[{"role": "system", "content": "count to 4"}],
         stream=True,
     )
 
-    for chunk in completion:
-        print(chunk)
+    merged = reduce(merge_openai_chunks, completion, None)
+    last_logged = last_logged_call()
+    assert (
+        last_logged.model_response.resp_payload["choices"][0]["message"]["content"]
+        == merged["choices"][0]["message"]["content"]
+    )
 
 
 async def test_async():
-    acompletion = await openai.ChatCompletion.acreate(
+    completion = await openai.ChatCompletion.acreate(
         model="gpt-3.5-turbo",
         messages=[{"role": "user", "content": "count down from 5"}],
     )
+    last_logged = last_logged_call()
+    assert (
+        last_logged.model_response.resp_payload["choices"][0]["message"]["content"]
+        == completion.choices[0].message.content
+    )
+    assert (
+        last_logged.model_response.req_payload["messages"][0]["content"]
+        == "count down from 5"
+    )
 
-    print(acompletion.choices[0].message.content)
+    assert completion.openpipe.cache_status == "SKIP"
 
 
 async def test_async_streaming():
-    acompletion = await openai.ChatCompletion.acreate(
+    completion = await openai.ChatCompletion.acreate(
         model="gpt-3.5-turbo",
         messages=[{"role": "user", "content": "count down from 5"}],
         stream=True,
     )
 
-    async for chunk in acompletion:
-        print(chunk)
+    merged = None
+    async for chunk in completion:
+        assert chunk.openpipe.cache_status == "SKIP"
+        merged = merge_openai_chunks(merged, chunk)
+
+    last_logged = last_logged_call()
+
+    assert (
+        last_logged.model_response.resp_payload["choices"][0]["message"]["content"]
+        == merged["choices"][0]["message"]["content"]
+    )
+    assert (
+        last_logged.model_response.req_payload["messages"][0]["content"]
+        == "count down from 5"
+    )
+    assert merged["openpipe"].cache_status == "SKIP"
 
 
 def test_sync_with_tags():
@@ -58,31 +109,54 @@ def test_sync_with_tags():
         messages=[{"role": "system", "content": "count to 10"}],
         openpipe={"tags": {"promptId": "testprompt"}},
     )
-    print("finished")
 
-    print(completion.choices[0].message.content)
+    last_logged = last_logged_call()
+    assert (
+        last_logged.model_response.resp_payload["choices"][0]["message"]["content"]
+        == completion.choices[0].message.content
+    )
+    print(last_logged.tags)
+    assert last_logged.tags["promptId"] == "testprompt"
+    assert last_logged.tags["$sdk"] == "python"
 
 
 def test_bad_call():
-    completion = openai.ChatCompletion.create(
-        model="gpt-3.5-turbo-blaster",
-        messages=[{"role": "system", "content": "count to 10"}],
-        stream=True,
+    try:
+        completion = openai.ChatCompletion.create(
+            model="gpt-3.5-turbo-blaster",
+            messages=[{"role": "system", "content": "count to 10"}],
+            stream=True,
+        )
+        assert False
+    except Exception as e:
+        pass
+    last_logged = last_logged_call()
+    print(last_logged)
+    assert (
+        last_logged.model_response.error_message
+        == "The model `gpt-3.5-turbo-blaster` does not exist"
     )
+    assert last_logged.model_response.status_code == 404
 
 
-@pytest.mark.focus
 async def test_caching():
+    messages = [{"role": "system", "content": f"repeat '{random_string(10)}'"}]
     completion = openai.ChatCompletion.create(
         model="gpt-3.5-turbo",
-        messages=[{"role": "system", "content": "count to 10"}],
+        messages=messages,
         openpipe={"cache": True},
     )
+    assert completion.openpipe.cache_status == "MISS"
+
+    first_logged = last_logged_call()
+    assert (
+        completion.choices[0].message.content
+        == first_logged.model_response.resp_payload["choices"][0]["message"]["content"]
+    )
 
     completion2 = await openai.ChatCompletion.acreate(
         model="gpt-3.5-turbo",
         messages=[{"role": "system", "content": "count to 10"}],
         openpipe={"cache": True},
     )
-
-    print(completion2)
+    assert completion2.openpipe.cache_status == "HIT"

client-libs/typescript/src/codegen/api.ts

@@ -4,7 +4,7 @@
  * OpenPipe API
  * The public API for reporting API calls to OpenPipe
  *
- * The version of the OpenAPI document: 0.1.0
+ * The version of the OpenAPI document: 0.1.1
  * 
  *
  * NOTE: This class is auto generated by OpenAPI Generator (https://openapi-generator.tech).
@@ -26,125 +26,193 @@ import { BASE_PATH, COLLECTION_FORMATS, BaseAPI, RequiredError } from './base';
 /**
  * 
  * @export
- * @interface ExternalApiCheckCache200Response
+ * @interface CheckCache200Response
  */
-export interface ExternalApiCheckCache200Response {
+export interface CheckCache200Response {
     /**
      * JSON-encoded response payload
      * @type {any}
-     * @memberof ExternalApiCheckCache200Response
+     * @memberof CheckCache200Response
      */
     'respPayload'?: any;
 }
 /**
  * 
  * @export
- * @interface ExternalApiCheckCacheDefaultResponse
+ * @interface CheckCacheDefaultResponse
  */
-export interface ExternalApiCheckCacheDefaultResponse {
+export interface CheckCacheDefaultResponse {
     /**
      * 
      * @type {string}
-     * @memberof ExternalApiCheckCacheDefaultResponse
+     * @memberof CheckCacheDefaultResponse
      */
     'message': string;
     /**
      * 
      * @type {string}
-     * @memberof ExternalApiCheckCacheDefaultResponse
+     * @memberof CheckCacheDefaultResponse
      */
     'code': string;
     /**
      * 
-     * @type {Array<ExternalApiCheckCacheDefaultResponseIssuesInner>}
-     * @memberof ExternalApiCheckCacheDefaultResponse
+     * @type {Array<CheckCacheDefaultResponseIssuesInner>}
+     * @memberof CheckCacheDefaultResponse
      */
-    'issues'?: Array<ExternalApiCheckCacheDefaultResponseIssuesInner>;
+    'issues'?: Array<CheckCacheDefaultResponseIssuesInner>;
 }
 /**
  * 
  * @export
- * @interface ExternalApiCheckCacheDefaultResponseIssuesInner
+ * @interface CheckCacheDefaultResponseIssuesInner
  */
-export interface ExternalApiCheckCacheDefaultResponseIssuesInner {
+export interface CheckCacheDefaultResponseIssuesInner {
     /**
      * 
      * @type {string}
-     * @memberof ExternalApiCheckCacheDefaultResponseIssuesInner
+     * @memberof CheckCacheDefaultResponseIssuesInner
      */
     'message': string;
 }
 /**
  * 
  * @export
- * @interface ExternalApiCheckCacheRequest
+ * @interface CheckCacheRequest
  */
-export interface ExternalApiCheckCacheRequest {
+export interface CheckCacheRequest {
     /**
      * Unix timestamp in milliseconds
      * @type {number}
-     * @memberof ExternalApiCheckCacheRequest
+     * @memberof CheckCacheRequest
      */
     'requestedAt': number;
     /**
      * JSON-encoded request payload
      * @type {any}
-     * @memberof ExternalApiCheckCacheRequest
+     * @memberof CheckCacheRequest
      */
     'reqPayload'?: any;
     /**
      * Extra tags to attach to the call for filtering. Eg { \"userId\": \"123\", \"promptId\": \"populate-title\" }
      * @type {{ [key: string]: string; }}
-     * @memberof ExternalApiCheckCacheRequest
+     * @memberof CheckCacheRequest
      */
     'tags'?: { [key: string]: string; };
 }
 /**
  * 
  * @export
- * @interface ExternalApiReportRequest
+ * @interface LocalTestingOnlyGetLatestLoggedCall200Response
  */
-export interface ExternalApiReportRequest {
+export interface LocalTestingOnlyGetLatestLoggedCall200Response {
+    /**
+     * 
+     * @type {string}
+     * @memberof LocalTestingOnlyGetLatestLoggedCall200Response
+     */
+    'createdAt': string;
+    /**
+     * 
+     * @type {boolean}
+     * @memberof LocalTestingOnlyGetLatestLoggedCall200Response
+     */
+    'cacheHit': boolean;
+    /**
+     * 
+     * @type {{ [key: string]: string; }}
+     * @memberof LocalTestingOnlyGetLatestLoggedCall200Response
+     */
+    'tags': { [key: string]: string; };
+    /**
+     * 
+     * @type {LocalTestingOnlyGetLatestLoggedCall200ResponseModelResponse}
+     * @memberof LocalTestingOnlyGetLatestLoggedCall200Response
+     */
+    'modelResponse': LocalTestingOnlyGetLatestLoggedCall200ResponseModelResponse | null;
+}
+/**
+ * 
+ * @export
+ * @interface LocalTestingOnlyGetLatestLoggedCall200ResponseModelResponse
+ */
+export interface LocalTestingOnlyGetLatestLoggedCall200ResponseModelResponse {
+    /**
+     * 
+     * @type {string}
+     * @memberof LocalTestingOnlyGetLatestLoggedCall200ResponseModelResponse
+     */
+    'id': string;
+    /**
+     * 
+     * @type {number}
+     * @memberof LocalTestingOnlyGetLatestLoggedCall200ResponseModelResponse
+     */
+    'statusCode': number | null;
+    /**
+     * 
+     * @type {string}
+     * @memberof LocalTestingOnlyGetLatestLoggedCall200ResponseModelResponse
+     */
+    'errorMessage': string | null;
+    /**
+     * 
+     * @type {any}
+     * @memberof LocalTestingOnlyGetLatestLoggedCall200ResponseModelResponse
+     */
+    'reqPayload'?: any;
+    /**
+     * 
+     * @type {any}
+     * @memberof LocalTestingOnlyGetLatestLoggedCall200ResponseModelResponse
+     */
+    'respPayload'?: any;
+}
+/**
+ * 
+ * @export
+ * @interface ReportRequest
+ */
+export interface ReportRequest {
     /**
      * Unix timestamp in milliseconds
      * @type {number}
-     * @memberof ExternalApiReportRequest
+     * @memberof ReportRequest
      */
     'requestedAt': number;
     /**
      * Unix timestamp in milliseconds
      * @type {number}
-     * @memberof ExternalApiReportRequest
+     * @memberof ReportRequest
      */
     'receivedAt': number;
     /**
      * JSON-encoded request payload
      * @type {any}
-     * @memberof ExternalApiReportRequest
+     * @memberof ReportRequest
      */
     'reqPayload'?: any;
     /**
      * JSON-encoded response payload
      * @type {any}
-     * @memberof ExternalApiReportRequest
+     * @memberof ReportRequest
      */
     'respPayload'?: any;
     /**
      * HTTP status code of response
      * @type {number}
-     * @memberof ExternalApiReportRequest
+     * @memberof ReportRequest
      */
     'statusCode'?: number;
     /**
      * User-friendly error message
      * @type {string}
-     * @memberof ExternalApiReportRequest
+     * @memberof ReportRequest
      */
     'errorMessage'?: string;
     /**
      * Extra tags to attach to the call for filtering. Eg { \"userId\": \"123\", \"promptId\": \"populate-title\" }
      * @type {{ [key: string]: string; }}
-     * @memberof ExternalApiReportRequest
+     * @memberof ReportRequest
      */
     'tags'?: { [key: string]: string; };
 }
@@ -157,14 +225,14 @@ export const DefaultApiAxiosParamCreator = function (configuration?: Configurati
     return {
         /**
          * Check if a prompt is cached
-         * @param {ExternalApiCheckCacheRequest} externalApiCheckCacheRequest 
+         * @param {CheckCacheRequest} checkCacheRequest 
          * @param {*} [options] Override http request option.
          * @throws {RequiredError}
          */
-        externalApiCheckCache: async (externalApiCheckCacheRequest: ExternalApiCheckCacheRequest, options: AxiosRequestConfig = {}): Promise<RequestArgs> => {
-            // verify required parameter 'externalApiCheckCacheRequest' is not null or undefined
-            assertParamExists('externalApiCheckCache', 'externalApiCheckCacheRequest', externalApiCheckCacheRequest)
-            const localVarPath = `/v1/check-cache`;
+        checkCache: async (checkCacheRequest: CheckCacheRequest, options: AxiosRequestConfig = {}): Promise<RequestArgs> => {
+            // verify required parameter 'checkCacheRequest' is not null or undefined
+            assertParamExists('checkCache', 'checkCacheRequest', checkCacheRequest)
+            const localVarPath = `/check-cache`;
             // use dummy base URL string because the URL constructor only accepts absolute URLs.
             const localVarUrlObj = new URL(localVarPath, DUMMY_BASE_URL);
             let baseOptions;
@@ -187,7 +255,40 @@ export const DefaultApiAxiosParamCreator = function (configuration?: Configurati
             setSearchParams(localVarUrlObj, localVarQueryParameter);
             let headersFromBaseOptions = baseOptions && baseOptions.headers ? baseOptions.headers : {};
             localVarRequestOptions.headers = {...localVarHeaderParameter, ...headersFromBaseOptions, ...options.headers};
-            localVarRequestOptions.data = serializeDataIfNeeded(externalApiCheckCacheRequest, localVarRequestOptions, configuration)
+            localVarRequestOptions.data = serializeDataIfNeeded(checkCacheRequest, localVarRequestOptions, configuration)
+
+            return {
+                url: toPathString(localVarUrlObj),
+                options: localVarRequestOptions,
+            };
+        },
+        /**
+         * Get the latest logged call (only for local testing)
+         * @param {*} [options] Override http request option.
+         * @throws {RequiredError}
+         */
+        localTestingOnlyGetLatestLoggedCall: async (options: AxiosRequestConfig = {}): Promise<RequestArgs> => {
+            const localVarPath = `/local-testing-only-get-latest-logged-call`;
+            // use dummy base URL string because the URL constructor only accepts absolute URLs.
+            const localVarUrlObj = new URL(localVarPath, DUMMY_BASE_URL);
+            let baseOptions;
+            if (configuration) {
+                baseOptions = configuration.baseOptions;
+            }
+
+            const localVarRequestOptions = { method: 'GET', ...baseOptions, ...options};
+            const localVarHeaderParameter = {} as any;
+            const localVarQueryParameter = {} as any;
+
+            // authentication Authorization required
+            // http bearer authentication required
+            await setBearerAuthToObject(localVarHeaderParameter, configuration)
+
+
+    
+            setSearchParams(localVarUrlObj, localVarQueryParameter);
+            let headersFromBaseOptions = baseOptions && baseOptions.headers ? baseOptions.headers : {};
+            localVarRequestOptions.headers = {...localVarHeaderParameter, ...headersFromBaseOptions, ...options.headers};
 
             return {
                 url: toPathString(localVarUrlObj),
@@ -196,14 +297,14 @@ export const DefaultApiAxiosParamCreator = function (configuration?: Configurati
         },
         /**
          * Report an API call
-         * @param {ExternalApiReportRequest} externalApiReportRequest 
+         * @param {ReportRequest} reportRequest 
          * @param {*} [options] Override http request option.
          * @throws {RequiredError}
          */
-        externalApiReport: async (externalApiReportRequest: ExternalApiReportRequest, options: AxiosRequestConfig = {}): Promise<RequestArgs> => {
-            // verify required parameter 'externalApiReportRequest' is not null or undefined
-            assertParamExists('externalApiReport', 'externalApiReportRequest', externalApiReportRequest)
-            const localVarPath = `/v1/report`;
+        report: async (reportRequest: ReportRequest, options: AxiosRequestConfig = {}): Promise<RequestArgs> => {
+            // verify required parameter 'reportRequest' is not null or undefined
+            assertParamExists('report', 'reportRequest', reportRequest)
+            const localVarPath = `/report`;
             // use dummy base URL string because the URL constructor only accepts absolute URLs.
             const localVarUrlObj = new URL(localVarPath, DUMMY_BASE_URL);
             let baseOptions;
@@ -226,7 +327,7 @@ export const DefaultApiAxiosParamCreator = function (configuration?: Configurati
             setSearchParams(localVarUrlObj, localVarQueryParameter);
             let headersFromBaseOptions = baseOptions && baseOptions.headers ? baseOptions.headers : {};
             localVarRequestOptions.headers = {...localVarHeaderParameter, ...headersFromBaseOptions, ...options.headers};
-            localVarRequestOptions.data = serializeDataIfNeeded(externalApiReportRequest, localVarRequestOptions, configuration)
+            localVarRequestOptions.data = serializeDataIfNeeded(reportRequest, localVarRequestOptions, configuration)
 
             return {
                 url: toPathString(localVarUrlObj),
@@ -245,22 +346,31 @@ export const DefaultApiFp = function(configuration?: Configuration) {
     return {
         /**
          * Check if a prompt is cached
-         * @param {ExternalApiCheckCacheRequest} externalApiCheckCacheRequest 
+         * @param {CheckCacheRequest} checkCacheRequest 
          * @param {*} [options] Override http request option.
          * @throws {RequiredError}
          */
-        async externalApiCheckCache(externalApiCheckCacheRequest: ExternalApiCheckCacheRequest, options?: AxiosRequestConfig): Promise<(axios?: AxiosInstance, basePath?: string) => AxiosPromise<ExternalApiCheckCache200Response>> {
-            const localVarAxiosArgs = await localVarAxiosParamCreator.externalApiCheckCache(externalApiCheckCacheRequest, options);
+        async checkCache(checkCacheRequest: CheckCacheRequest, options?: AxiosRequestConfig): Promise<(axios?: AxiosInstance, basePath?: string) => AxiosPromise<CheckCache200Response>> {
+            const localVarAxiosArgs = await localVarAxiosParamCreator.checkCache(checkCacheRequest, options);
+            return createRequestFunction(localVarAxiosArgs, globalAxios, BASE_PATH, configuration);
+        },
+        /**
+         * Get the latest logged call (only for local testing)
+         * @param {*} [options] Override http request option.
+         * @throws {RequiredError}
+         */
+        async localTestingOnlyGetLatestLoggedCall(options?: AxiosRequestConfig): Promise<(axios?: AxiosInstance, basePath?: string) => AxiosPromise<LocalTestingOnlyGetLatestLoggedCall200Response>> {
+            const localVarAxiosArgs = await localVarAxiosParamCreator.localTestingOnlyGetLatestLoggedCall(options);
             return createRequestFunction(localVarAxiosArgs, globalAxios, BASE_PATH, configuration);
         },
         /**
          * Report an API call
-         * @param {ExternalApiReportRequest} externalApiReportRequest 
+         * @param {ReportRequest} reportRequest 
          * @param {*} [options] Override http request option.
          * @throws {RequiredError}
          */
-        async externalApiReport(externalApiReportRequest: ExternalApiReportRequest, options?: AxiosRequestConfig): Promise<(axios?: AxiosInstance, basePath?: string) => AxiosPromise<any>> {
-            const localVarAxiosArgs = await localVarAxiosParamCreator.externalApiReport(externalApiReportRequest, options);
+        async report(reportRequest: ReportRequest, options?: AxiosRequestConfig): Promise<(axios?: AxiosInstance, basePath?: string) => AxiosPromise<any>> {
+            const localVarAxiosArgs = await localVarAxiosParamCreator.report(reportRequest, options);
             return createRequestFunction(localVarAxiosArgs, globalAxios, BASE_PATH, configuration);
         },
     }
@@ -275,21 +385,29 @@ export const DefaultApiFactory = function (configuration?: Configuration, basePa
     return {
         /**
          * Check if a prompt is cached
-         * @param {ExternalApiCheckCacheRequest} externalApiCheckCacheRequest 
+         * @param {CheckCacheRequest} checkCacheRequest 
          * @param {*} [options] Override http request option.
          * @throws {RequiredError}
          */
-        externalApiCheckCache(externalApiCheckCacheRequest: ExternalApiCheckCacheRequest, options?: any): AxiosPromise<ExternalApiCheckCache200Response> {
-            return localVarFp.externalApiCheckCache(externalApiCheckCacheRequest, options).then((request) => request(axios, basePath));
+        checkCache(checkCacheRequest: CheckCacheRequest, options?: any): AxiosPromise<CheckCache200Response> {
+            return localVarFp.checkCache(checkCacheRequest, options).then((request) => request(axios, basePath));
+        },
+        /**
+         * Get the latest logged call (only for local testing)
+         * @param {*} [options] Override http request option.
+         * @throws {RequiredError}
+         */
+        localTestingOnlyGetLatestLoggedCall(options?: any): AxiosPromise<LocalTestingOnlyGetLatestLoggedCall200Response> {
+            return localVarFp.localTestingOnlyGetLatestLoggedCall(options).then((request) => request(axios, basePath));
         },
         /**
          * Report an API call
-         * @param {ExternalApiReportRequest} externalApiReportRequest 
+         * @param {ReportRequest} reportRequest 
          * @param {*} [options] Override http request option.
          * @throws {RequiredError}
          */
-        externalApiReport(externalApiReportRequest: ExternalApiReportRequest, options?: any): AxiosPromise<any> {
-            return localVarFp.externalApiReport(externalApiReportRequest, options).then((request) => request(axios, basePath));
+        report(reportRequest: ReportRequest, options?: any): AxiosPromise<any> {
+            return localVarFp.report(reportRequest, options).then((request) => request(axios, basePath));
         },
     };
 };
@@ -303,24 +421,34 @@ export const DefaultApiFactory = function (configuration?: Configuration, basePa
 export class DefaultApi extends BaseAPI {
     /**
      * Check if a prompt is cached
-     * @param {ExternalApiCheckCacheRequest} externalApiCheckCacheRequest 
+     * @param {CheckCacheRequest} checkCacheRequest 
      * @param {*} [options] Override http request option.
      * @throws {RequiredError}
      * @memberof DefaultApi
      */
-    public externalApiCheckCache(externalApiCheckCacheRequest: ExternalApiCheckCacheRequest, options?: AxiosRequestConfig) {
-        return DefaultApiFp(this.configuration).externalApiCheckCache(externalApiCheckCacheRequest, options).then((request) => request(this.axios, this.basePath));
+    public checkCache(checkCacheRequest: CheckCacheRequest, options?: AxiosRequestConfig) {
+        return DefaultApiFp(this.configuration).checkCache(checkCacheRequest, options).then((request) => request(this.axios, this.basePath));
+    }
+
+    /**
+     * Get the latest logged call (only for local testing)
+     * @param {*} [options] Override http request option.
+     * @throws {RequiredError}
+     * @memberof DefaultApi
+     */
+    public localTestingOnlyGetLatestLoggedCall(options?: AxiosRequestConfig) {
+        return DefaultApiFp(this.configuration).localTestingOnlyGetLatestLoggedCall(options).then((request) => request(this.axios, this.basePath));
     }
 
     /**
      * Report an API call
-     * @param {ExternalApiReportRequest} externalApiReportRequest 
+     * @param {ReportRequest} reportRequest 
      * @param {*} [options] Override http request option.
      * @throws {RequiredError}
      * @memberof DefaultApi
      */
-    public externalApiReport(externalApiReportRequest: ExternalApiReportRequest, options?: AxiosRequestConfig) {
-        return DefaultApiFp(this.configuration).externalApiReport(externalApiReportRequest, options).then((request) => request(this.axios, this.basePath));
+    public report(reportRequest: ReportRequest, options?: AxiosRequestConfig) {
+        return DefaultApiFp(this.configuration).report(reportRequest, options).then((request) => request(this.axios, this.basePath));
     }
 }
 

client-libs/typescript/src/codegen/base.ts

@@ -4,7 +4,7 @@
  * OpenPipe API
  * The public API for reporting API calls to OpenPipe
  *
- * The version of the OpenAPI document: 0.1.0
+ * The version of the OpenAPI document: 0.1.1
  * 
  *
  * NOTE: This class is auto generated by OpenAPI Generator (https://openapi-generator.tech).
@@ -19,7 +19,7 @@ import type { Configuration } from './configuration';
 import type { AxiosPromise, AxiosInstance, AxiosRequestConfig } from 'axios';
 import globalAxios from 'axios';
 
-export const BASE_PATH = "https://app.openpipe.ai/api".replace(/\/+$/, "");
+export const BASE_PATH = "https://app.openpipe.ai/api/v1".replace(/\/+$/, "");
 
 /**
  *

client-libs/typescript/src/codegen/common.ts

@@ -4,7 +4,7 @@
  * OpenPipe API
  * The public API for reporting API calls to OpenPipe
  *
- * The version of the OpenAPI document: 0.1.0
+ * The version of the OpenAPI document: 0.1.1
  * 
  *
  * NOTE: This class is auto generated by OpenAPI Generator (https://openapi-generator.tech).

client-libs/typescript/src/codegen/configuration.ts

@@ -4,7 +4,7 @@
  * OpenPipe API
  * The public API for reporting API calls to OpenPipe
  *
- * The version of the OpenAPI document: 0.1.0
+ * The version of the OpenAPI document: 0.1.1
  * 
  *
  * NOTE: This class is auto generated by OpenAPI Generator (https://openapi-generator.tech).

client-libs/typescript/src/codegen/index.ts

@@ -4,7 +4,7 @@
  * OpenPipe API
  * The public API for reporting API calls to OpenPipe
  *
- * The version of the OpenAPI document: 0.1.0
+ * The version of the OpenAPI document: 0.1.1
  * 
  *
  * NOTE: This class is auto generated by OpenAPI Generator (https://openapi-generator.tech).