@earendil-works/pi-ai 1.0.4 → 1.2.0

package/README.md

@@ -60,7 +60,7 @@ Unified LLM API with provider collections, automatic auth resolution, token and
 
 ## Supported Providers
 
-- **OpenAI**
+- **OpenAI** (including the Decisions classifier API)
 - **Ant Ling**
 - **Azure OpenAI (Responses)**
 - **OpenAI Codex (legacy)** (ChatGPT Plus/Pro subscription, requires OAuth, see below)
@@ -890,7 +890,7 @@ console.log(model.output); // ['image'] or ['image', 'text']
 
 ## Classification
 
-Classifier models consume structured JSON state and answer one or more typed questions. They do not use chat or image-generation APIs. TypeSafe's Jev model is available from these built-in providers:
+Classifier models consume structured JSON state, and optionally images, and answer one or more typed questions. They do not use chat or image-generation APIs. TypeSafe's Jev model, Cloudflare's Clef models, and OpenAI's GPT-6 Luna are available from these built-in providers:
 
 | Provider | Model IDs | Auth |
 | --- | --- | --- |
@@ -899,6 +899,7 @@ Classifier models consume structured JSON state and answer one or more typed que
 | `cloudflare-workers-ai` | `typesafe/jev`, `@cf/cloudflare/clef`, `@cf/cloudflare/clef-flash` | `CLOUDFLARE_API_KEY` and `CLOUDFLARE_ACCOUNT_ID` |
 | `vercel-ai-gateway` | `typesafe-ai/jev` | `AI_GATEWAY_API_KEY` |
 | `opencode` | `jev-1.13`, `jev-1.13-free` | `OPENCODE_API_KEY` |
+| `openai` | `gpt-6-luna` | `OPENAI_API_KEY` |
 
 ```typescript
 import { builtinModels } from '@earendil-works/pi-ai/providers/all';
@@ -936,10 +937,20 @@ The public contract uses `bool` questions and `{ type: "bool", probability }` an
 
 When the service reports token counts, `result.usage` carries them with their cost at the model's catalog price, the same `Usage` shape as chat messages. All System One services report token counts; a request that was answered with malformed answers keeps its usage. Local classifiers such as `llama-cpp-classify` report no usage.
 
-`ClassifierOptions.temperature` divides the answer logits by the given value before they are normalized; values above 1 soften the distribution. APIs that cannot apply it, such as System One, ignore it.
+`ClassifierOptions.temperature` divides the answer logits by the given value before they are normalized; values above 1 soften the distribution. APIs that cannot apply it, such as System One and OpenAI Decisions, ignore it.
+
+`ClassifierContext.images` holds image blocks (`{ type: "image", data, mimeType }`) judged together with the state. Only models whose `input` includes `"image"` accept them; `classify()` returns an error result for other models.
+
+### OpenAI Decisions
+
+The `openai-decisions` API calls OpenAI's [Decisions API](https://developers.openai.com/api/docs/guides/decisions) (`POST /v1/decisions`). GPT-6 Luna is its only model. The state is sent as JSON text; with images, the input is one user message with the state followed by the images as data URLs (at most 128). `choice` and `score` questions map to the Decisions types of the same name. `bool` questions become `predicate` questions; predicates have no criteria field, so the meanings of true and false are appended to the instructions. If OpenAI refuses to answer a question, the whole result is an error. Usage reports input tokens only, priced at the catalog input rate with OpenAI's long-context multiplier.
+
+The endpoint needs an OpenAI API key; Sign in with ChatGPT credentials are rejected, so the `openai` provider hides `gpt-6-luna` from `getAvailableOfType('classifier')` while it uses OAuth. The API rejects inputs above 922K tokens, but requests that run longer than about five seconds, currently above roughly 600K input tokens, fail with a gateway timeout (504).
 
 ### Chat models on llama.cpp
 
+llama.cpp also serves decision models such as Julia-1 or Kev natively through `/v1/systemone`. Since llama.cpp 0.6.0, `GET /models` lists `decisions` in a model's `architecture.output_modalities` for these models. They use the `typesafe-system-one` API with the server's `/v1` URL as `baseUrl` (for example `http://127.0.0.1:8080/v1`); it needs an `apiKey`, which llama.cpp ignores unless it was started with `--api-key`. The `llama-cpp-classify` API is the fallback for chat models.
+
 The `llama-cpp-classify` API turns a chat model served by llama.cpp's `llama-server` into a classifier. Each question becomes one chat prompt: the state, every question of the request, the state again, and the question with its answers under single-token labels (letters for a choice, `Yes`/`No` for a bool, digits for a score). The prompt up to the final question is shared by all questions of a request, so the server's prompt cache evaluates the state once per request. The server returns the log-probabilities of the next token, and the answer is the softmax over the label tokens. Choices support up to 62 options and scores up to 10 levels. The model's `baseUrl` is the server URL; a trailing `/v1` is ignored. In router mode, the model ID selects the model.
 
 ```typescript

package/dist/api/anthropic-messages.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"anthropic-messages.d.ts","sourceRoot":"","sources":["../../src/api/anthropic-messages.ts"],"names":[],"mappings":"AAAA,OAAO,SAAiC,MAAM,mBAAmB,CAAC;AAoBlE,OAAO,KAAK,EASX,mBAAmB,EAEnB,cAAc,EACd,aAAa,EAMb,MAAM,aAAa,CAAC;AAmJrB,MAAM,MAAM,eAAe,GAAG,KAAK,GAAG,QAAQ,GAAG,MAAM,GAAG,OAAO,GAAG,KAAK,CAAC;AAE1E,MAAM,MAAM,wBAAwB,GAAG,YAAY,GAAG,SAAS,CAAC;AA2ChE,MAAM,WAAW,gBAAiB,SAAQ,aAAa;IACtD;;;;;;OAMG;IACH,eAAe,CAAC,EAAE,OAAO,CAAC;IAC1B;;;;OAIG;IACH,oBAAoB,CAAC,EAAE,MAAM,CAAC;IAC9B;;;;;;;;;;;OAWG;IACH,MAAM,CAAC,EAAE,eAAe,CAAC;IACzB;;;;;;;;;;;OAWG;IACH,eAAe,CAAC,EAAE,wBAAwB,CAAC;IAC3C;;;;;OAKG;IACH,mBAAmB,CAAC,EAAE,OAAO,CAAC;IAC9B;;;;OAIG;IACH,UAAU,CAAC,EAAE,MAAM,GAAG,KAAK,GAAG,MAAM,GAAG;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC;IACtE;;;;OAIG;IACH,MAAM,CAAC,EAAE,SAAS,CAAC;CACnB;AAwRD,eAAO,MAAM,MAAM,EAAE,cAAc,CAAC,oBAAoB,EAAE,gBAAgB,CA0UzE,CAAC;AA2BF,eAAO,MAAM,YAAY,EAAE,cAAc,CAAC,oBAAoB,EAAE,mBAAmB,CAgDlF,CAAC"}
+{"version":3,"file":"anthropic-messages.d.ts","sourceRoot":"","sources":["../../src/api/anthropic-messages.ts"],"names":[],"mappings":"AAAA,OAAO,SAAiC,MAAM,mBAAmB,CAAC;AAoBlE,OAAO,KAAK,EASX,mBAAmB,EAEnB,cAAc,EACd,aAAa,EAMb,MAAM,aAAa,CAAC;AAmJrB,MAAM,MAAM,eAAe,GAAG,KAAK,GAAG,QAAQ,GAAG,MAAM,GAAG,OAAO,GAAG,KAAK,CAAC;AAE1E,MAAM,MAAM,wBAAwB,GAAG,YAAY,GAAG,SAAS,CAAC;AA2ChE,MAAM,WAAW,gBAAiB,SAAQ,aAAa;IACtD;;;;;;OAMG;IACH,eAAe,CAAC,EAAE,OAAO,CAAC;IAC1B;;;;OAIG;IACH,oBAAoB,CAAC,EAAE,MAAM,CAAC;IAC9B;;;;;;;;;;;OAWG;IACH,MAAM,CAAC,EAAE,eAAe,CAAC;IACzB;;;;;;;;;;;OAWG;IACH,eAAe,CAAC,EAAE,wBAAwB,CAAC;IAC3C;;;;;OAKG;IACH,mBAAmB,CAAC,EAAE,OAAO,CAAC;IAC9B;;;;OAIG;IACH,UAAU,CAAC,EAAE,MAAM,GAAG,KAAK,GAAG,MAAM,GAAG;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC;IACtE;;;;OAIG;IACH,MAAM,CAAC,EAAE,SAAS,CAAC;CACnB;AAwRD,eAAO,MAAM,MAAM,EAAE,cAAc,CAAC,oBAAoB,EAAE,gBAAgB,CA2WzE,CAAC;AA2BF,eAAO,MAAM,YAAY,EAAE,cAAc,CAAC,oBAAoB,EAAE,mBAAmB,CAgDlF,CAAC"}

package/dist/api/anthropic-messages.js

@@ -425,21 +425,53 @@ export const stream = (model, context, options) => {
                 client = created.client;
                 isOAuth = created.isOAuthToken;
             }
-            let params = buildParams(model, normalizedContext, isOAuth, options);
-            const nextParams = await options?.onPayload?.(params, model);
-            if (nextParams !== undefined) {
-                params = { ...nextParams, stream: true };
-            }
             const requestOptions = {
                 ...(options?.signal ? { signal: options.signal } : {}),
                 ...(options?.timeoutMs !== undefined ? { timeout: options.timeoutMs } : {}),
                 maxRetries: 0,
             };
-            const response = await retryProviderRequest(() => client.beta.messages.create(params, requestOptions).asResponse(), {
-                maxRetries: options?.maxRetries,
-                maxRetryDelayMs: options?.maxRetryDelayMs,
-                signal: options?.signal,
-            });
+            let disablePreferredStrictTools = false;
+            let response;
+            for (;;) {
+                options?.signal?.throwIfAborted();
+                const built = buildParams(model, normalizedContext, isOAuth, options, disablePreferredStrictTools);
+                let params = built.params;
+                const nextParams = await options?.onPayload?.(params, model);
+                if (nextParams !== undefined) {
+                    params = { ...nextParams, stream: true };
+                }
+                try {
+                    response = await retryProviderRequest(() => client.beta.messages.create(params, requestOptions).asResponse(), {
+                        maxRetries: options?.maxRetries,
+                        maxRetryDelayMs: options?.maxRetryDelayMs,
+                        signal: options?.signal,
+                    });
+                    break;
+                }
+                catch (error) {
+                    // Only downgrade a rejected request, never an already-started stream.
+                    if (options?.signal?.aborted ||
+                        disablePreferredStrictTools ||
+                        !built.hasPreferredStrictTools ||
+                        !(error instanceof Anthropic.APIError) ||
+                        error.status !== 400 ||
+                        error.type !== "invalid_request_error" ||
+                        ![
+                            "The compiled grammar is too large",
+                            "Too many strict tools (",
+                            "Schemas contains too many parameters with union types (",
+                        ].some((message) => error.message.includes(message))) {
+                        throw error;
+                    }
+                    disablePreferredStrictTools = true;
+                    appendAssistantMessageDiagnostic(output, {
+                        type: "anthropic_strict_tools_fallback",
+                        timestamp: Date.now(),
+                        error: { message: error.message },
+                        details: { reason: "strict_tool_limits_exceeded" },
+                    });
+                }
+            }
             await options?.onResponse?.({ status: response.status, headers: headersToRecord(response.headers) }, model);
             stream.push({ type: "start", partial: output });
             const blocks = output.content;
@@ -855,7 +887,7 @@ function getBetaFeatures(model, context, isOAuthToken, nativeToolChanges, option
         features.push(INLINE_TOOLS_BETA);
     return [...new Set(features)];
 }
-function buildParams(model, context, isOAuthToken, options) {
+function buildParams(model, context, isOAuthToken, options, disablePreferredStrictTools = false) {
     const { cacheControl } = getCacheControl(model, options?.cacheRetention, options?.env);
     const compat = getAnthropicCompat(model);
     const initialSystemMessage = getInitialSystemMessage(context.messages);
@@ -866,11 +898,20 @@ function buildParams(model, context, isOAuthToken, options) {
     // by value in a `tool_addition` block, which also expresses same-name redefinitions.
     // Anthropic rejects a tool list where every tool is deferred, so there must be an initial
     // active tool to anchor the placeholder. Otherwise the current tool list is sent.
+    let hasPreferredStrictTools = false;
+    const convertToolDefinitions = (tools, toolCacheControl) => {
+        const definitions = convertTools(tools, isOAuthToken, compat.supportsEagerToolInputStreaming, compat.supportsStrictTools, toolCacheControl, disablePreferredStrictTools);
+        if (definitions.some((definition, index) => {
+            const config = tools[index].constrainedSampling;
+            return definition.strict === true && config && config.type === "json_schema" && config.strict === "prefer";
+        })) {
+            hasPreferredStrictTools = true;
+        }
+        return definitions;
+    };
     const initialTools = initialSystemMessage?.toolsAdded ?? [];
     const nativeToolChanges = compat.supportsMidConvoSystemMessages && compat.supportsMidConvoToolChanges && initialTools.length > 0;
-    const converted = convertMessages(conversationMessages, isOAuthToken, cacheControl, compat.allowEmptySignature, model.compat?.supportsMidConvoEffort === true ? model.provider : undefined, nativeToolChanges
-        ? (tools) => convertTools(tools, isOAuthToken, compat.supportsEagerToolInputStreaming, compat.supportsStrictTools)
-        : undefined);
+    const converted = convertMessages(conversationMessages, isOAuthToken, cacheControl, compat.allowEmptySignature, model.compat?.supportsMidConvoEffort === true ? model.provider : undefined, nativeToolChanges ? convertToolDefinitions : undefined);
     const activeEffort = options?.effort ?? "high";
     const betaFeatures = getBetaFeatures(model, context, isOAuthToken, nativeToolChanges, options);
     const params = {
@@ -922,15 +963,12 @@ function buildParams(model, context, isOAuthToken, options) {
         // placeholder. The list never changes afterwards: later tools are defined by value in
         // `tool_addition` blocks and withdrawn by `tool_removal`, so the cached prefix survives
         // every tool change.
-        params.tools = [
-            ...convertTools(initialTools, isOAuthToken, compat.supportsEagerToolInputStreaming, compat.supportsStrictTools, toolCacheControl),
-            DEFERRED_TOOL_PLACEHOLDER,
-        ];
+        params.tools = [...convertToolDefinitions(initialTools, toolCacheControl), DEFERRED_TOOL_PLACEHOLDER];
     }
     else {
         const tools = getCurrentTools(context.messages);
         if (tools.length > 0) {
-            params.tools = convertTools(tools, isOAuthToken, compat.supportsEagerToolInputStreaming, compat.supportsStrictTools, toolCacheControl);
+            params.tools = convertToolDefinitions(tools, toolCacheControl);
         }
     }
     // Managed effort models always use adaptive thinking so prefix mismatches can
@@ -986,7 +1024,7 @@ function buildParams(model, context, isOAuthToken, options) {
     if (allowedFallbackModels && allowedFallbackModels.length > 0) {
         params.fallbacks = allowedFallbackModels.map((fallback) => ({ model: fallback.model }));
     }
-    return params;
+    return { params, hasPreferredStrictTools };
 }
 // Normalize tool call IDs to match Anthropic's required pattern and length
 function normalizeToolCallId(id) {
@@ -1255,11 +1293,14 @@ const isAnthropicStrictUnsupportedKeyword = (key, value) => {
         return typeof value !== "string" || !ANTHROPIC_STRICT_STRING_FORMATS.has(value);
     return false;
 };
-function convertTools(tools, isOAuthToken, supportsEagerToolInputStreaming, supportsStrictTools, cacheControl) {
+function convertTools(tools, isOAuthToken, supportsEagerToolInputStreaming, supportsStrictTools, cacheControl, disablePreferredStrictTools = false) {
     if (!tools)
         return [];
     return tools.map((tool, index) => {
-        const strict = resolveJsonSchemaStrictSampling(tool, supportsStrictTools, isAnthropicStrictUnsupportedKeyword);
+        const config = tool.constrainedSampling;
+        const strict = disablePreferredStrictTools && config && config.type === "json_schema" && config.strict === "prefer"
+            ? undefined
+            : resolveJsonSchemaStrictSampling(tool, supportsStrictTools, isAnthropicStrictUnsupportedKeyword);
         const parameters = getJsonSchemaToolParameters(tool, strict);
         const schema = parameters;
         const legacyInputSchema = {