promptopskit 0.9.1 → 0.10.0

package/README.md

@@ -219,14 +219,16 @@ result = await kit.renderPrompt({
 });
 if (!result.request) throw new Error(result.returnMessage ?? 'Prompt rendering failed.');
 
-// LLMAsAService — OpenAI-compatible gateway with project and customer metadata
+// LLMAsAService — OpenAI-compatible gateway with bearer auth and customer metadata
 result = await kit.renderPrompt({
   path: 'hello',
   provider: 'llmasaservice',
+  llmasaservice: {
+    apiKey: process.env.LLM_GATEWAY_API_KEY!,
+  },
   runtime: {
     provider_options: {
       llmasaservice: {
-        project_id: process.env.LLM_GATEWAY_PROJECT_ID,
         customer: { customer_id: 'cust_123', customer_name: 'Acme' },
       },
     },
@@ -235,7 +237,7 @@ result = await kit.renderPrompt({
 });
 if (!result.request) throw new Error(result.returnMessage ?? 'Prompt rendering failed.');
 // result.request.body → { model, messages, customer, ... }
-// result.request.headers → { 'x-project-id': '...' }
+// result.request.headers → { Authorization: 'Bearer ...' }
 ```
 
 Provider adapters are also available as direct imports:
@@ -292,7 +294,7 @@ cache:
     retention: 24h
 ```
 
-At render time, pass the caller-owned API key. PromptOpsKit calls `POST https://api.thetokencompany.com/v1/compress` directly with `fetch`; no TheTokenCompany SDK is required.
+At render time, pass the caller-owned API key. PromptOpsKit calls `POST https://api.thetokencompany.com/v1/compress` directly with `fetch`; no TheTokenCompany SDK is required. If credentials are unavailable or the backend call fails, PromptOpsKit preserves the original prompt text, returns a `POK057` warning in `warnings`, and reports zero token savings with matching input/output token counts. Library rendering does not log this fallback to the console.
 
 ```typescript
 const result = await kit.renderPrompt({
@@ -341,7 +343,7 @@ TOON preprocessing uses a local encode-only implementation inspired by the MIT-l
 
 See [Compression and Compaction](./docs/compression.md) for complete examples and token-savings reporting.
 
-Use `provider_options` when PromptOpsKit has a known provider-specific mapping, such as Anthropic `top_k`, Gemini's native `response_schema`, OpenRouter routing fields, or LLMAsAService gateway routing/customer metadata.
+Use `provider_options` when PromptOpsKit has a known provider-specific mapping, such as Anthropic `top_k`, Gemini's native `response_schema`, OpenRouter routing fields, or LLMAsAService customer/conversation metadata.
 
 ```yaml
 response:
@@ -359,14 +361,13 @@ provider_options:
       order: ["anthropic", "openai"]
     transforms: ["middle-out"]
   llmasaservice:
-    project_id: "llm-project-id"
     # Optional default; usually pass the real customer at render time.
     customer:
       customer_id: "cust_123"
       customer_name: "Acme"
 ```
 
-For LLMAsAService, `provider_options.llmasaservice.customer` is intended to be render-time attribution for the current account/user. A prompt can keep a default, but production calls should normally override it through `runtime.provider_options.llmasaservice.customer`.
+For LLMAsAService, pass the gateway credential through the render-time `llmasaservice.apiKey` option; the adapter emits it as an `Authorization: Bearer ...` header. `provider_options.llmasaservice.customer` is intended to be render-time attribution for the current account/user. A prompt can keep a default, but production calls should normally override it through `runtime.provider_options.llmasaservice.customer`. A project id is no longer required.
 
 When a provider adds a body field PromptOpsKit does not model yet, use `raw`:
 
@@ -689,6 +690,7 @@ Renders a prompt for a specific provider. Returns `{ resolved, request?, returnM
 | `toolRegistry` | `Record<string, unknown>` | Tool definitions for resolving string tool references |
 | `strict` | `boolean` | Fail on missing variables except object-form inputs marked `optional: true` |
 | `openaiResponses` | `object` | Optional Responses API extras (`previous_response_id`, `conversation`, `instructions`, `parallel_tool_calls`, `max_tool_calls`, `store`, `metadata`, `include`, `background`) |
+| `llmasaservice` | `object` | Required LLMAsAService gateway credentials (`apiKey`) when using that provider |
 | `theTokenCompany` | `object` | Optional compression settings (`apiKey`, `baseURL`, `fetch`) used when `compression.thetokencompany.enabled: true` |
 
 Use `compressionSummary.tokensSaved` for a lightweight operation-level aggregate across compression and compaction steps. Use `compression` for the detailed per-step breakdown.

package/SKILL.md

@@ -269,8 +269,8 @@ import {
 } from 'promptopskit/llmasaservice';
 
 const client = new OpenAI(createLLMAsAServiceOpenAIConfig({
+  apiKey: process.env.LLM_GATEWAY_API_KEY!,
   baseURL: process.env.LLM_GATEWAY_BASE_URL,
-  projectId: process.env.LLM_GATEWAY_PROJECT_ID,
 }));
 
 const result = await llmasaserviceAdapter.renderPrompt(
@@ -278,10 +278,12 @@ const result = await llmasaserviceAdapter.renderPrompt(
     path: 'support/triage-summary',
   },
   {
+    llmasaservice: {
+      apiKey: process.env.LLM_GATEWAY_API_KEY!,
+    },
     runtime: {
       provider_options: {
         llmasaservice: {
-          project_id: process.env.LLM_GATEWAY_PROJECT_ID,
           customer: {
             customer_id: customer.id,
             customer_name: customer.name,
@@ -772,14 +774,13 @@ provider_options:
       order: ["anthropic", "openai"]
     transforms: ["middle-out"]
   llmasaservice:
-    project_id: llm-project-id
     # Optional default; usually pass the real customer at render time.
     customer:
       customer_id: cust_123
       customer_name: Acme
 ```
 
-For LLMAsAService, prefer putting the current customer/account/user attribution in `runtime.provider_options.llmasaservice.customer` during rendering. Static prompt metadata may include a default, but runtime values should override it for real requests.
+For LLMAsAService, pass the server-owned gateway credential through `llmasaservice.apiKey`; the adapter emits `Authorization: Bearer <api-key>`. Prefer putting the current customer/account/user attribution in `runtime.provider_options.llmasaservice.customer` during rendering. Static prompt metadata may include a default, but runtime values should override it for real requests. Do not require or generate a project id.
 
 Use `raw` only when a vendor request-body field is important and PromptOpsKit does not model it yet:
 

package/dist/{chunk-QZHR2YMK.js → chunk-46Z64RVO.js}

@@ -1,7 +1,7 @@
 import {
   loadPromptFile,
   parsePrompt
-} from "./chunk-VOXMOGU5.js";
+} from "./chunk-ALBSFXYR.js";
 
 // src/renderer/interpolate.ts
 var VARIABLE_RE = /\{\{\s*([a-zA-Z_][a-zA-Z0-9_]*)(?:\s*\|\s*(compress|toon|compact|code))?\s*\}\}/g;
@@ -1632,6 +1632,7 @@ async function applyPromptCompressionForRender(asset, runtime) {
       aggressiveness: theTokenCompanyConfig?.aggressiveness
     });
     promptTemplate = result.output;
+    warnings.push(...result.warnings ?? []);
     compression.push({
       provider: "thetokencompany",
       model,
@@ -1697,41 +1698,80 @@ function reportHeuristicCompressionWarnings2(warnings, compressionWarnings, scop
 async function compressWithTheTokenCompany(input, options) {
   const apiKey = options.apiKey ?? getEnv("THETOKENCOMPANY_API_KEY") ?? getEnv("TTC_API_KEY");
   if (!apiKey) {
-    throw new Error(
-      "TheTokenCompany compression is enabled, but no API key was provided. Pass theTokenCompany.apiKey to renderPrompt() or set THETOKENCOMPANY_API_KEY."
+    return createTheTokenCompanyFallback(
+      input,
+      "no API key was provided"
     );
   }
   const fetchImpl = options.fetch ?? globalThis.fetch;
   if (!fetchImpl) {
-    throw new Error("TheTokenCompany compression requires a runtime with fetch support.");
+    return createTheTokenCompanyFallback(
+      input,
+      "fetch is unavailable in this runtime"
+    );
   }
   const baseURL = (options.baseURL ?? THETOKENCOMPANY_DEFAULT_BASE_URL).replace(/\/+$/, "");
   const compressionSettings = options.aggressiveness === void 0 ? void 0 : { aggressiveness: options.aggressiveness };
-  const response = await fetchImpl(`${baseURL}/v1/compress`, {
-    method: "POST",
-    headers: {
-      Authorization: `Bearer ${apiKey}`,
-      "Content-Type": "application/json"
-    },
-    body: JSON.stringify({
-      model: options.model,
+  let response;
+  try {
+    response = await fetchImpl(`${baseURL}/v1/compress`, {
+      method: "POST",
+      headers: {
+        Authorization: `Bearer ${apiKey}`,
+        "Content-Type": "application/json"
+      },
+      body: JSON.stringify({
+        model: options.model,
+        input,
+        ...compressionSettings ? { compression_settings: compressionSettings } : {}
+      })
+    });
+  } catch (error) {
+    return createTheTokenCompanyFallback(
       input,
-      ...compressionSettings ? { compression_settings: compressionSettings } : {}
-    })
-  });
+      `request failed: ${toErrorMessage(error)}`
+    );
+  }
   if (!response.ok) {
-    const body = await response.text().catch(() => "");
-    throw new Error(
-      `TheTokenCompany compression failed with HTTP ${response.status}` + (body ? `: ${body}` : ".")
+    return createTheTokenCompanyFallback(
+      input,
+      `service returned HTTP ${response.status}`
+    );
+  }
+  let data;
+  try {
+    data = await response.json();
+  } catch (error) {
+    return createTheTokenCompanyFallback(
+      input,
+      `response body was not valid JSON: ${toErrorMessage(error)}`
     );
   }
-  const data = await response.json();
   const normalized = normalizeTheTokenCompanyCompressResponse(data);
   if (!normalized) {
-    throw new Error("TheTokenCompany compression returned an invalid response payload.");
+    return createTheTokenCompanyFallback(
+      input,
+      "response payload was invalid"
+    );
   }
   return normalized;
 }
+function createTheTokenCompanyFallback(input, reason) {
+  const tokens = estimateHeuristicTokens(input);
+  return {
+    output: input,
+    output_tokens: tokens,
+    input_tokens: tokens,
+    tokens_saved: 0,
+    compression_ratio: tokens === 0 ? 0 : 1,
+    warnings: [
+      `POK057: TheTokenCompany compression skipped; using uncompressed prompt with zero token savings (${reason}).`
+    ]
+  };
+}
+function toErrorMessage(error) {
+  return error instanceof Error && error.message ? error.message : "unknown error";
+}
 function normalizeTheTokenCompanyCompressResponse(data) {
   if (typeof data.output !== "string" || typeof data.output_tokens !== "number") {
     return void 0;
@@ -1857,4 +1897,4 @@ export {
   withPromptInputSupport,
   applyRawProviderBody
 };
-//# sourceMappingURL=chunk-QZHR2YMK.js.map
+//# sourceMappingURL=chunk-46Z64RVO.js.map