@tensorcost/sdk 0.3.0

package/CHANGELOG.md

@@ -0,0 +1,36 @@
+# Changelog
+
+## 0.3.0 — 2026-05-12
+
+- Added: `appliedMode` config flag. When true and `proxyUrl` (or `TENSORCOST_PROXY_URL`) is set, requests route through the TensorCost inference-proxy. See `docs/features/inference-proxy-applied-mode.md`.
+- Added: `TensorCostConfigError` — thrown at `wrap()` time when `appliedMode=true` and no proxy URL is configured.
+- AWS Bedrock clients throw immediately at `wrap()` time when `appliedMode=true` (v1 limitation — SigV4 signing is not interceptable in the same pattern; observe-only mode still works).
+- No breaking changes to the 0.2.x observe-only API.
+
+## 0.2.1 — 2026-05-05
+
+Phase A4 batch 4 — `connectionId` support. Stamps `connection_id` on every observation envelope when the SDK is initialised with `wrap(client, { connectionId })` or `TENSORCOST_CONNECTION_ID`. The backend uses this as a fallback for environment resolution when the `environment` field is absent.
+
+## 0.1.1 — 2026-05-05
+
+### Fixed
+- Observation transport now POSTs to `/api/inference-proxy/observation`
+  (and exchanges tokens at `/api/inference-proxy/sdk-token/exchange`),
+  matching the inference-proxy-service controller mount per the
+  `/api/<svc>/<feature>` routing convention. The 0.1.0 paths
+  (`/api/proxy/...`) returned 404 against every TensorCost deployment,
+  so customer-side observations never landed.
+
+## 0.1.0 — 2026-05-05
+
+Initial release.
+
+- `wrap(client, options?)` one-line entry point.
+- OpenAI provider — wraps `chat.completions.create` and `completions.create`.
+- Anthropic provider — wraps `messages.create`.
+- Fire-and-forget HTTP transport over native `fetch` (Node 18+).
+- Short-lived JWT exchange against `POST /api/inference-proxy/sdk-token/exchange` (the 0.1.0 paste-up said `/api/proxy/...`; corrected to the real route in 0.1.1), cached + auto-refreshed.
+- Bounded queue (1024) with drop-on-saturation backpressure.
+- Fail-open by default. The single exception: missing `apiKey` throws `MissingConfigError` at `wrap()` time.
+- Zero runtime dependencies.
+- TypeScript-first; ships with `.d.ts`.

package/LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for describing the origin of the Work and
+      reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2026 Vaadh Labs, Inc. (TensorCost)
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

package/README.md

@@ -0,0 +1,119 @@
+# @tensorcost/sdk
+
+One-line wrapper SDK for TensorCost — fire-and-forget LLM observability for OpenAI and Anthropic clients in Node.js.
+
+## Install
+
+```bash
+npm install @tensorcost/sdk
+# or
+pnpm add @tensorcost/sdk
+```
+
+Requires Node.js 18+ (uses native `fetch`).
+
+## Use
+
+```ts
+import OpenAI from "openai";
+import { wrap } from "@tensorcost/sdk";
+
+const client = wrap(new OpenAI({ apiKey: process.env.OPENAI_API_KEY }));
+
+// Use the client exactly as before — TensorCost observes every call
+// and ships fire-and-forget observations to the platform.
+const resp = await client.chat.completions.create({
+  model: "gpt-4o-mini",
+  messages: [{ role: "user", content: "hi" }],
+});
+```
+
+Anthropic is just as simple:
+
+```ts
+import Anthropic from "@anthropic-ai/sdk";
+import { wrap } from "@tensorcost/sdk";
+
+const client = wrap(new Anthropic({ apiKey: process.env.ANTHROPIC_API_KEY }));
+
+await client.messages.create({
+  model: "claude-3-5-sonnet-20241022",
+  max_tokens: 256,
+  messages: [{ role: "user", content: "hi" }],
+});
+```
+
+## Configuration
+
+`wrap()` reads from the environment by default:
+
+| Variable | Purpose | Required |
+|---|---|---|
+| `TENSORCOST_API_KEY` | Long-lived API key from the TensorCost console | Yes |
+| `TENSORCOST_BASE_URL` | Backend base URL | No (default `https://api.tensorcost.com`) |
+| `TENSORCOST_TENANT_ID` | Optional explicit tenant id | No |
+| `TENSORCOST_PROXY_URL` | Proxy base URL for applied mode | Required when `appliedMode: true` |
+| `TENSORCOST_ENVIRONMENT` | Environment tag stamped on every observation | No |
+| `TENSORCOST_CONNECTION_ID` | Provider-connection identifier | No |
+
+Or pass explicitly:
+
+```ts
+const client = wrap(new OpenAI(...), {
+  apiKey: "tc_live_xxx",
+  baseUrl: "https://api.tensorcost.com",
+});
+```
+
+## Applied mode (Layer 2)
+
+When you enable applied mode, the SDK routes inference requests through the TensorCost inference-proxy instead of directly to the AI provider. The proxy decides per-request whether to re-route to an alternate provider or pass through unchanged — the SDK doesn't make that call.
+
+```ts
+const client = wrap(new OpenAI({ apiKey: process.env.OPENAI_API_KEY }), {
+  apiKey: "tc_live_xxx",
+  appliedMode: true,
+  proxyUrl: "https://proxy.tensorcost.com", // or set TENSORCOST_PROXY_URL
+});
+
+// Use normally — the SDK redirects the client to the proxy transparently.
+// Observe-only telemetry still fires after every call.
+const resp = await client.chat.completions.create({
+  model: "gpt-4o-mini",
+  messages: [{ role: "user", content: "hi" }],
+});
+```
+
+`wrap()` throws `TensorCostConfigError` at import time if `appliedMode: true` and neither `proxyUrl` nor `TENSORCOST_PROXY_URL` is set. AWS Bedrock is not yet supported in applied mode (SigV4 signing can't be intercepted in the same pattern); observe-only mode works fine for Bedrock.
+
+See [`docs/features/inference-proxy-applied-mode.md`](../../docs/features/inference-proxy-applied-mode.md) for the proxy-side design.
+
+## Fail-open guarantee
+
+If TensorCost is slow, down, or misconfigured, your customer call STILL succeeds. The SDK swallows internal errors and surfaces them via `console.warn`. The one exception: a missing API key throws `MissingConfigError` at `wrap()` time — silently dropping every observation would be worse than a loud import-time failure.
+
+## What is captured
+
+Only metadata — never prompt or completion content:
+
+- Provider + model
+- Operation (`chat.completions`, `completions`, `messages`)
+- Request / response timestamps
+- Input / output token counts (from the provider's `usage` field)
+- Status (`success` / `error`) and error class/message on failure
+- A correlation UUID
+
+## Currently supported
+
+- OpenAI (`openai` >= 4.x) — `chat.completions.create`, `completions.create`
+- Anthropic (`@anthropic-ai/sdk` >= 0.x) — `messages.create`
+
+## Coming soon
+
+- Bedrock, Azure OpenAI, Vertex
+- Streaming responses (`stream: true`)
+- Tool / function calling (multi-call trace stitching)
+
+## License
+
+Proprietary — Vaadh Labs / TensorCost.

package/dist/applied.d.ts

@@ -0,0 +1,98 @@
+/**
+ * Applied-mode wiring — Layer 2.
+ *
+ * When the customer opts into applied mode, the SDK routes every inference
+ * request through the TensorCost inference-proxy instead of directly to the
+ * upstream AI provider. The proxy decides per-request whether to re-route
+ * or pass through unchanged; the SDK doesn't make that call.
+ *
+ * The mechanism is straightforward: at `wrap()` time we read the
+ * provider's original baseURL and auth from the client, permanently
+ * redirect the client's baseURL to the proxy endpoint, and inject two
+ * custom headers the proxy expects:
+ *
+ *   x-tc-provider-url   — the original provider base URL
+ *   x-tc-provider-auth  — the auth header value to forward upstream
+ *
+ * Both the OpenAI SDK and the Anthropic SDK expose `baseURL` and
+ * `defaultHeaders` as first-class properties, so we can set them
+ * without monkey-patching the HTTP layer.
+ *
+ * Bedrock is not supported in v1. The AWS SDK uses SigV4 signing on
+ * regional endpoints and doesn't expose baseURL / defaultHeaders in
+ * the same shape. Passing a Bedrock-shaped client with appliedMode=true
+ * throws `TensorCostConfigError` at wrap time.
+ *
+ * Observe-only telemetry (the 0.2.x substrate) continues to fire
+ * after every call regardless of this setting.
+ */
+/**
+ * Client shapes we understand. All detection is duck-typed; no import
+ * of the actual openai or @anthropic-ai/sdk packages.
+ */
+/** Header value the proxy expects for OpenAI + Azure OpenAI. */
+declare function bearerAuth(apiKey: string): string;
+/** Header value the proxy expects for Anthropic. */
+declare function anthropicAuth(apiKey: string): string;
+/**
+ * Read the original baseURL from a duck-typed client. Both the OpenAI
+ * and Anthropic SDKs expose a `baseURL` string property on the client
+ * instance (the actual HTTP origin, e.g. "https://api.openai.com/v1").
+ * Returns null when the property is absent or not a string.
+ */
+declare function readBaseUrl(client: Record<string, unknown>): string | null;
+/**
+ * Read the API key from the client. OpenAI SDK exposes `client.apiKey`;
+ * Anthropic SDK exposes `client.apiKey` too. Returns null when absent.
+ */
+declare function readApiKey(client: Record<string, unknown>): string | null;
+/**
+ * Detect whether this client looks like an Azure OpenAI client (the
+ * openai package's `AzureOpenAI` subclass). Azure uses `api-key` header
+ * auth and a custom endpoint URL, so the auth value we forward differs
+ * from vanilla OpenAI.
+ *
+ * Detection: Azure clients have an `azure` property set to true, OR
+ * their baseURL contains `.openai.azure.com`.
+ */
+declare function isAzureOpenAI(client: Record<string, unknown>): boolean;
+/**
+ * Azure OpenAI uses an `api-key` header rather than `Authorization: Bearer`.
+ * The proxy forwards `x-tc-provider-auth` verbatim as the upstream auth
+ * header, so we need to format it as the proxy expects: "api-key <value>".
+ *
+ * Note: The proxy currently only supports `Authorization` and `api-key`
+ * style headers. This value is forwarded in the `x-tc-provider-auth`
+ * header and then sent by the proxy as the upstream Authorization.
+ */
+declare function azureAuth(apiKey: string): string;
+/**
+ * Install applied-mode routing on the client. Called once by `wrap()`.
+ *
+ * Mutates `client.baseURL` to point at the proxy endpoint and injects
+ * `x-tc-provider-url` / `x-tc-provider-auth` into `client.defaultHeaders`.
+ * These headers survive for the lifetime of the client — every subsequent
+ * request automatically carries them.
+ *
+ * @param client   - The customer's AI SDK client (duck-typed).
+ * @param proxyUrl - The TensorCost proxy base URL (already stripped of trailing slash).
+ * @param provider - "openai" | "anthropic", as detected by detectProvider().
+ *
+ * @throws TensorCostConfigError
+ *   - If the client doesn't expose `baseURL` (can't determine the original
+ *     provider URL to forward in `x-tc-provider-url`).
+ *   - If the client doesn't expose `apiKey` (can't build `x-tc-provider-auth`).
+ *   - If the provider is "bedrock" (not yet supported in v1).
+ */
+export declare function installAppliedMode(client: unknown, proxyUrl: string, provider: "openai" | "anthropic"): void;
+export declare const PROXY_PATH_SUFFIX = "/api/inference-proxy/v1/chat/completions";
+export declare const __test__: {
+    readBaseUrl: typeof readBaseUrl;
+    readApiKey: typeof readApiKey;
+    isAzureOpenAI: typeof isAzureOpenAI;
+    bearerAuth: typeof bearerAuth;
+    anthropicAuth: typeof anthropicAuth;
+    azureAuth: typeof azureAuth;
+};
+export {};
+//# sourceMappingURL=applied.d.ts.map

package/dist/applied.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"applied.d.ts","sourceRoot":"","sources":["../src/applied.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;GA2BG;AAOH;;;GAGG;AAEH,gEAAgE;AAChE,iBAAS,UAAU,CAAC,MAAM,EAAE,MAAM,GAAG,MAAM,CAE1C;AAED,oDAAoD;AACpD,iBAAS,aAAa,CAAC,MAAM,EAAE,MAAM,GAAG,MAAM,CAE7C;AAED;;;;;GAKG;AACH,iBAAS,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,MAAM,GAAG,IAAI,CAMnE;AAED;;;GAGG;AACH,iBAAS,UAAU,CAAC,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,MAAM,GAAG,IAAI,CAGlE;AAED;;;;;;;;GAQG;AACH,iBAAS,aAAa,CAAC,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,OAAO,CAI/D;AAED;;;;;;;;GAQG;AACH,iBAAS,SAAS,CAAC,MAAM,EAAE,MAAM,GAAG,MAAM,CAEzC;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,kBAAkB,CAChC,MAAM,EAAE,OAAO,EACf,QAAQ,EAAE,MAAM,EAChB,QAAQ,EAAE,QAAQ,GAAG,WAAW,GAC/B,IAAI,CAiEN;AAED,eAAO,MAAM,iBAAiB,6CAAa,CAAC;AAE5C,eAAO,MAAM,QAAQ;;;;;;;CAOpB,CAAC"}

package/dist/applied.js

@@ -0,0 +1,174 @@
+/**
+ * Applied-mode wiring — Layer 2.
+ *
+ * When the customer opts into applied mode, the SDK routes every inference
+ * request through the TensorCost inference-proxy instead of directly to the
+ * upstream AI provider. The proxy decides per-request whether to re-route
+ * or pass through unchanged; the SDK doesn't make that call.
+ *
+ * The mechanism is straightforward: at `wrap()` time we read the
+ * provider's original baseURL and auth from the client, permanently
+ * redirect the client's baseURL to the proxy endpoint, and inject two
+ * custom headers the proxy expects:
+ *
+ *   x-tc-provider-url   — the original provider base URL
+ *   x-tc-provider-auth  — the auth header value to forward upstream
+ *
+ * Both the OpenAI SDK and the Anthropic SDK expose `baseURL` and
+ * `defaultHeaders` as first-class properties, so we can set them
+ * without monkey-patching the HTTP layer.
+ *
+ * Bedrock is not supported in v1. The AWS SDK uses SigV4 signing on
+ * regional endpoints and doesn't expose baseURL / defaultHeaders in
+ * the same shape. Passing a Bedrock-shaped client with appliedMode=true
+ * throws `TensorCostConfigError` at wrap time.
+ *
+ * Observe-only telemetry (the 0.2.x substrate) continues to fire
+ * after every call regardless of this setting.
+ */
+import { TensorCostConfigError } from "./config.js";
+/** Proxy chat-completions path — fixed in v1. */
+const PROXY_PATH = "/api/inference-proxy/v1/chat/completions";
+/**
+ * Client shapes we understand. All detection is duck-typed; no import
+ * of the actual openai or @anthropic-ai/sdk packages.
+ */
+/** Header value the proxy expects for OpenAI + Azure OpenAI. */
+function bearerAuth(apiKey) {
+    return `Bearer ${apiKey}`;
+}
+/** Header value the proxy expects for Anthropic. */
+function anthropicAuth(apiKey) {
+    return `x-api-key ${apiKey}`;
+}
+/**
+ * Read the original baseURL from a duck-typed client. Both the OpenAI
+ * and Anthropic SDKs expose a `baseURL` string property on the client
+ * instance (the actual HTTP origin, e.g. "https://api.openai.com/v1").
+ * Returns null when the property is absent or not a string.
+ */
+function readBaseUrl(client) {
+    const raw = client["baseURL"];
+    if (typeof raw === "string" && raw.length > 0) {
+        return raw.replace(/\/+$/, "");
+    }
+    return null;
+}
+/**
+ * Read the API key from the client. OpenAI SDK exposes `client.apiKey`;
+ * Anthropic SDK exposes `client.apiKey` too. Returns null when absent.
+ */
+function readApiKey(client) {
+    const raw = client["apiKey"];
+    return typeof raw === "string" && raw.length > 0 ? raw : null;
+}
+/**
+ * Detect whether this client looks like an Azure OpenAI client (the
+ * openai package's `AzureOpenAI` subclass). Azure uses `api-key` header
+ * auth and a custom endpoint URL, so the auth value we forward differs
+ * from vanilla OpenAI.
+ *
+ * Detection: Azure clients have an `azure` property set to true, OR
+ * their baseURL contains `.openai.azure.com`.
+ */
+function isAzureOpenAI(client) {
+    if (client["azure"] === true)
+        return true;
+    const url = readBaseUrl(client) ?? "";
+    return url.includes(".openai.azure.com");
+}
+/**
+ * Azure OpenAI uses an `api-key` header rather than `Authorization: Bearer`.
+ * The proxy forwards `x-tc-provider-auth` verbatim as the upstream auth
+ * header, so we need to format it as the proxy expects: "api-key <value>".
+ *
+ * Note: The proxy currently only supports `Authorization` and `api-key`
+ * style headers. This value is forwarded in the `x-tc-provider-auth`
+ * header and then sent by the proxy as the upstream Authorization.
+ */
+function azureAuth(apiKey) {
+    return `api-key ${apiKey}`;
+}
+/**
+ * Install applied-mode routing on the client. Called once by `wrap()`.
+ *
+ * Mutates `client.baseURL` to point at the proxy endpoint and injects
+ * `x-tc-provider-url` / `x-tc-provider-auth` into `client.defaultHeaders`.
+ * These headers survive for the lifetime of the client — every subsequent
+ * request automatically carries them.
+ *
+ * @param client   - The customer's AI SDK client (duck-typed).
+ * @param proxyUrl - The TensorCost proxy base URL (already stripped of trailing slash).
+ * @param provider - "openai" | "anthropic", as detected by detectProvider().
+ *
+ * @throws TensorCostConfigError
+ *   - If the client doesn't expose `baseURL` (can't determine the original
+ *     provider URL to forward in `x-tc-provider-url`).
+ *   - If the client doesn't expose `apiKey` (can't build `x-tc-provider-auth`).
+ *   - If the provider is "bedrock" (not yet supported in v1).
+ */
+export function installAppliedMode(client, proxyUrl, provider) {
+    if (provider === "bedrock") {
+        // Bedrock uses SigV4 signing on regional HTTPS endpoints. The AWS
+        // SDK does not expose baseURL / defaultHeaders in the same shape,
+        // so we can't intercept the request path in v1. Pass-through-only
+        // mode is the correct posture; applied mode for Bedrock is a
+        // follow-up workstream.
+        throw new TensorCostConfigError("TensorCost applied mode is not yet supported for AWS Bedrock clients. " +
+            "Set appliedMode: false to continue with observe-only mode, or use " +
+            "an OpenAI / Anthropic client.");
+    }
+    const c = client;
+    const originalBaseUrl = readBaseUrl(c);
+    if (!originalBaseUrl) {
+        throw new TensorCostConfigError("TensorCost applied mode requires a client with a readable `baseURL` " +
+            "property (the original provider endpoint). The client passed to " +
+            "wrap() does not expose one.");
+    }
+    const apiKey = readApiKey(c);
+    if (!apiKey) {
+        throw new TensorCostConfigError("TensorCost applied mode requires a client with a readable `apiKey` " +
+            "property to forward provider auth. The client passed to wrap() does " +
+            "not expose one.");
+    }
+    // Build the auth header value to forward. The proxy sends this value
+    // as the upstream Authorization (or api-key) header verbatim.
+    let providerAuth;
+    if (provider === "anthropic") {
+        providerAuth = anthropicAuth(apiKey);
+    }
+    else if (isAzureOpenAI(c)) {
+        providerAuth = azureAuth(apiKey);
+    }
+    else {
+        providerAuth = bearerAuth(apiKey);
+    }
+    // Redirect the client's base URL to the proxy. The path suffix
+    // (/api/inference-proxy/v1) is appended here so that the OpenAI-compatible
+    // client doesn't double-prepend its own /v1 path. Both the OpenAI and
+    // Anthropic SDKs append the method path to baseURL; the proxy endpoint is
+    // OpenAI-compatible, so the client will build:
+    //   baseURL + "/chat/completions"
+    // We set baseURL to the proxy's parent path to get the right URL.
+    const cleanProxyUrl = proxyUrl.replace(/\/+$/, "");
+    c["baseURL"] = `${cleanProxyUrl}/api/inference-proxy/v1`;
+    // Inject the custom headers into the client's defaultHeaders object.
+    // Both SDKs merge defaultHeaders with every outgoing request.
+    const existing = c["defaultHeaders"];
+    const headers = existing && typeof existing === "object" && !Array.isArray(existing)
+        ? existing
+        : {};
+    headers["x-tc-provider-url"] = originalBaseUrl;
+    headers["x-tc-provider-auth"] = providerAuth;
+    c["defaultHeaders"] = headers;
+}
+export const PROXY_PATH_SUFFIX = PROXY_PATH;
+export const __test__ = {
+    readBaseUrl,
+    readApiKey,
+    isAzureOpenAI,
+    bearerAuth,
+    anthropicAuth,
+    azureAuth,
+};
+//# sourceMappingURL=applied.js.map

package/dist/applied.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"applied.js","sourceRoot":"","sources":["../src/applied.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;GA2BG;AAEH,OAAO,EAAE,qBAAqB,EAAE,MAAM,aAAa,CAAC;AAEpD,iDAAiD;AACjD,MAAM,UAAU,GAAG,0CAA0C,CAAC;AAE9D;;;GAGG;AAEH,gEAAgE;AAChE,SAAS,UAAU,CAAC,MAAc;IAChC,OAAO,UAAU,MAAM,EAAE,CAAC;AAC5B,CAAC;AAED,oDAAoD;AACpD,SAAS,aAAa,CAAC,MAAc;IACnC,OAAO,aAAa,MAAM,EAAE,CAAC;AAC/B,CAAC;AAED;;;;;GAKG;AACH,SAAS,WAAW,CAAC,MAA+B;IAClD,MAAM,GAAG,GAAG,MAAM,CAAC,SAAS,CAAC,CAAC;IAC9B,IAAI,OAAO,GAAG,KAAK,QAAQ,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC9C,OAAO,GAAG,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC;IACjC,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED;;;GAGG;AACH,SAAS,UAAU,CAAC,MAA+B;IACjD,MAAM,GAAG,GAAG,MAAM,CAAC,QAAQ,CAAC,CAAC;IAC7B,OAAO,OAAO,GAAG,KAAK,QAAQ,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC;AAChE,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,aAAa,CAAC,MAA+B;IACpD,IAAI,MAAM,CAAC,OAAO,CAAC,KAAK,IAAI;QAAE,OAAO,IAAI,CAAC;IAC1C,MAAM,GAAG,GAAG,WAAW,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;IACtC,OAAO,GAAG,CAAC,QAAQ,CAAC,mBAAmB,CAAC,CAAC;AAC3C,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,SAAS,CAAC,MAAc;IAC/B,OAAO,WAAW,MAAM,EAAE,CAAC;AAC7B,CAAC;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,UAAU,kBAAkB,CAChC,MAAe,EACf,QAAgB,EAChB,QAAgC;IAEhC,IAAI,QAAkB,KAAK,SAAS,EAAE,CAAC;QACrC,kEAAkE;QAClE,kEAAkE;QAClE,kEAAkE;QAClE,6DAA6D;QAC7D,wBAAwB;QACxB,MAAM,IAAI,qBAAqB,CAC7B,wEAAwE;YACtE,oEAAoE;YACpE,+BAA+B,CAClC,CAAC;IACJ,CAAC;IAED,MAAM,CAAC,GAAG,MAAiC,CAAC;IAE5C,MAAM,eAAe,GAAG,WAAW,CAAC,CAAC,CAAC,CAAC;IACvC,IAAI,CAAC,eAAe,EAAE,CAAC;QACrB,MAAM,IAAI,qBAAqB,CAC7B,sEAAsE;YACpE,kEAAkE;YAClE,6BAA6B,CAChC,CAAC;IACJ,CAAC;IAED,MAAM,MAAM,GAAG,UAAU,CAAC,CAAC,CAAC,CAAC;IAC7B,IAAI,CAAC,MAAM,EAAE,CAAC;QACZ,MAAM,IAAI,qBAAqB,CAC7B,qEAAqE;YACnE,sEAAsE;YACtE,iBAAiB,CACpB,CAAC;IACJ,CAAC;IAED,qEAAqE;IACrE,8DAA8D;IAC9D,IAAI,YAAoB,CAAC;IACzB,IAAI,QAAQ,KAAK,WAAW,EAAE,CAAC;QAC7B,YAAY,GAAG,aAAa,CAAC,MAAM,CAAC,CAAC;IACvC,CAAC;SAAM,IAAI,aAAa,CAAC,CAAC,CAAC,EAAE,CAAC;QAC5B,YAAY,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC;IACnC,CAAC;SAAM,CAAC;QACN,YAAY,GAAG,UAAU,CAAC,MAAM,CAAC,CAAC;IACpC,CAAC;IAED,+DAA+D;IAC/D,2EAA2E;IAC3E,sEAAsE;IACtE,0EAA0E;IAC1E,+CAA+C;IAC/C,kCAAkC;IAClC,kEAAkE;IAClE,MAAM,aAAa,GAAG,QAAQ,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC;IACnD,CAAC,CAAC,SAAS,CAAC,GAAG,GAAG,aAAa,yBAAyB,CAAC;IAEzD,qEAAqE;IACrE,8DAA8D;IAC9D,MAAM,QAAQ,GAAG,CAAC,CAAC,gBAAgB,CAAC,CAAC;IACrC,MAAM,OAAO,GACX,QAAQ,IAAI,OAAO,QAAQ,KAAK,QAAQ,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC;QAClE,CAAC,CAAE,QAAmC;QACtC,CAAC,CAAC,EAAE,CAAC;IACT,OAAO,CAAC,mBAAmB,CAAC,GAAG,eAAe,CAAC;IAC/C,OAAO,CAAC,oBAAoB,CAAC,GAAG,YAAY,CAAC;IAC7C,CAAC,CAAC,gBAAgB,CAAC,GAAG,OAAO,CAAC;AAChC,CAAC;AAED,MAAM,CAAC,MAAM,iBAAiB,GAAG,UAAU,CAAC;AAE5C,MAAM,CAAC,MAAM,QAAQ,GAAG;IACtB,WAAW;IACX,UAAU;IACV,aAAa;IACb,UAAU;IACV,aAAa;IACb,SAAS;CACV,CAAC"}

package/dist/config.d.ts

@@ -0,0 +1,29 @@
+/**
+ * Configuration resolution for the SDK.
+ *
+ * Resolution order:
+ *   1. Explicit options to wrap()
+ *   2. Environment variables (TENSORCOST_API_KEY, TENSORCOST_BASE_URL,
+ *      TENSORCOST_TENANT_ID, TENSORCOST_PROXY_URL)
+ *   3. Defaults (only for baseUrl)
+ *
+ * If apiKey is still missing after resolution, MissingConfigError is
+ * raised. This is the *one* case the SDK does not fail open — it means
+ * the SDK was never configured at all, and silently dropping every
+ * observation would be worse than a loud import-time failure.
+ *
+ * If appliedMode=true and proxyUrl is missing from both options and env,
+ * TensorCostConfigError is thrown. Applied mode needs an explicit proxy
+ * URL; there is no safe default.
+ */
+import type { ResolvedConfig, WrapOptions } from "./types.js";
+export declare const DEFAULT_BASE_URL = "https://api.tensorcost.com";
+export declare class MissingConfigError extends Error {
+    constructor(message: string);
+}
+/** Thrown when applied mode is enabled but required config is absent. */
+export declare class TensorCostConfigError extends Error {
+    constructor(message: string);
+}
+export declare function resolveConfig(options?: WrapOptions): ResolvedConfig;
+//# sourceMappingURL=config.d.ts.map