promptopskit 0.3.8 → 0.4.0

package/README.md

@@ -11,6 +11,31 @@ Your prompts are already in Git. PromptOpsKit makes them manageable. It replaces
 
 Provider adapters for OpenAI, Anthropic, Gemini, and OpenRouter produce a ready-to-send **request body only** — no HTTP client, no auth, no headers. Your application owns transport, so PromptOpsKit slots into any stack without opinions about how you call the API.
 
+## How is this different from GitHub Models?
+
+GitHub Models is a great place to prototype prompts, compare models, and run evaluations inside GitHub.
+
+PromptOpsKit is focused on the application runtime layer.
+
+It helps you turn prompt behavior into repo-native assets that your application can load, validate, compose, compile, and render into provider-specific request bodies.
+
+Use GitHub Models when you want:
+
+- a GitHub-hosted prompt editor and playground
+- side-by-side model comparisons
+- structured evaluations inside GitHub
+- `.prompt.yml` files for prompt experiments and evals
+
+Use PromptOpsKit when you want:
+
+- Markdown prompt assets with YAML metadata
+- production input hardening and validation
+- environment-specific model and parameter overrides
+- reusable prompt composition
+- compiled artifacts for deployment
+- provider-specific request bodies for your own runtime code
+- control over SDKs, auth, retries, routing, observability, and billing
+
 ### Why PromptOpsKit?
 
 - **Centralized, not scattered** — each prompt is a single Markdown file that captures prompt text, model config, tool bindings, and context rules together.
@@ -44,9 +69,12 @@ This creates:
 prompts/
 ├── defaults.md         # Folder-level defaults (provider, model, metadata, system instructions)
 ├── hello.md            # Sample prompt with variables
-├── hello.test.yaml     # Test sidecar with sample inputs
+├── hello.test.yaml     # Test sidecar with sample inputs and hardcoded responses
 └── shared/
     └── tone.md         # Shared system instructions (included via composition)
+
+tests/
+└── hello.prompt.test.mjs # Executable starter test for the hello prompt
 ```
 
 ### 2. Write a prompt
@@ -69,9 +97,7 @@ context:
       reject_secrets: true
     - name: app_context
       max_size: 2000
-      allow_regex:
-        pattern: "^[A-Za-z0-9 _-]+$"
-        flags: "i"
+      allow_regex: /^[A-Za-z0-9 _-]+$/i
 includes:
   - ./shared/tone.md
 ---
@@ -135,9 +161,11 @@ Supported values for `warnings.contextSize` are `auto`, `off`, `result-only`, `c
 - **Composition** — `includes` to share system instructions across prompts, with circular detection
 - **Folder defaults** — `defaults.md` inheritance for shared provider, model, metadata, and system instructions
 - **Overrides** — Environment and tier-based overrides (base → env → tier → runtime)
-- **4 provider adapters** — OpenAI, Anthropic, Gemini, OpenRouter — body-only output
+- **5 provider adapters** — OpenAI (Chat), OpenAI (Responses), Anthropic, Gemini, OpenRouter — body-only output
+- **Provider-aware input caching controls** — optional `cache` front matter maps to OpenAI prompt cache hints, Anthropic `cache_control`, and Gemini `cachedContent`
+- **Vendor escape hatch** — optional `raw.<provider>` blocks shallow-merge unmodeled request-body fields into the final provider payload
 - **Validation** — Zod schema validation, Levenshtein-based "did you mean?" for typos, variable usage checks
-- **Context hardening** — structured regexes with flags, `/pattern/i` convenience syntax, and built-in `non_empty` / `reject_secrets` validators
+- **Context hardening** — copyable `/pattern/i` regex literals, structured regexes with `return_message`, and built-in `non_empty` / `reject_secrets` validators
 - **Optional short-circuit messages** — validators can return a structured `returnMessage` instead of throwing when configured
 - **Context size guardrails** — optional per-input `max_size` metadata with non-blocking render-time warnings
 - **Warning controls** — top-level config can suppress or emit context size warnings differently in dev and prod
@@ -193,6 +221,7 @@ Provider adapters are also available as direct imports:
 
 ```typescript
 import { openaiAdapter } from 'promptopskit/openai';
+import { openaiResponsesAdapter } from 'promptopskit/openai-responses';
 import { anthropicAdapter } from 'promptopskit/anthropic';
 import { geminiAdapter } from 'promptopskit/gemini';
 import { openrouterAdapter } from 'promptopskit/openrouter';
@@ -223,6 +252,48 @@ const request = openaiAdapter.render(prompt, {
 
 In browser or client-side code, keep provider credentials on the server. Use the rendered request body with your own server endpoint, server action, or edge function rather than calling a provider directly from the client.
 
+### Provider-specific fields and raw passthrough
+
+Use normalized fields first (`sampling`, `response`, `cache`, `tools`) so prompts stay portable. `response.schema` is the neutral JSON Schema path; adapters emit it as OpenAI/OpenRouter `response_format`, OpenAI Responses `text.format`, Anthropic `output_config.format`, and Gemini `generationConfig.responseJsonSchema`.
+
+Use `provider_options` when PromptOpsKit has a known provider-specific mapping, such as Anthropic `top_k`, Gemini's native `response_schema`, or OpenRouter routing fields.
+
+```yaml
+response:
+  format: json
+  schema_name: support_reply
+  schema_description: Structured support reply
+  schema:
+    type: object
+    properties:
+      answer:
+        type: string
+provider_options:
+  openrouter:
+    provider:
+      order: ["anthropic", "openai"]
+    transforms: ["middle-out"]
+```
+
+When a provider adds a body field PromptOpsKit does not model yet, use `raw`:
+
+```yaml
+raw:
+  openai:
+    service_tier: flex
+  anthropic:
+    service_tier: auto
+  gemini:
+    safetySettings:
+      - category: HARM_CATEGORY_DANGEROUS_CONTENT
+        threshold: BLOCK_ONLY_HIGH
+  openrouter:
+    usage:
+      include: true
+```
+
+Each adapter reads only its matching raw block and shallow-merges it into the generated request body after normalized mappings. This is intentionally an escape hatch; prefer first-class fields when they exist.
+
 On the server, adapters also provide async prompt-aware helpers so you can use the default `./prompts` and `./.generated-prompts/json` directories without creating a `PromptOpsKit` instance:
 
 ```typescript
@@ -472,11 +543,21 @@ Hello {{ name }}!`,
 ## Testing Helpers
 
 ```typescript
-import { createMockAsset, createMockResolvedAsset, parseTestPrompt } from 'promptopskit/testing';
+import {
+  createHardcodedPromptResponder,
+  createMockAsset,
+  createMockResolvedAsset,
+  loadPromptTestSidecar,
+  parseTestPrompt,
+} from 'promptopskit/testing';
 
 const asset = createMockAsset({ model: 'gpt-5.4' });
 const resolved = createMockResolvedAsset();
 const parsed = parseTestPrompt('---\nid: test\nschema_version: 1\n---\n\nHello');
+
+const sidecar = await loadPromptTestSidecar('./prompts/hello.test.yaml');
+const respond = createHardcodedPromptResponder(sidecar);
+const response = respond('basic-greeting');
 ```
 
 ## API Reference
@@ -501,7 +582,7 @@ Renders a prompt for a specific provider. Returns `{ resolved, request?, returnM
 |--------|------|-------------|
 | `path` | `string` | Prompt path (no extension), e.g. `'support/reply'` |
 | `source` | `string` | Inline prompt source (alternative to path) |
-| `provider` | `string` | `'openai'`, `'anthropic'`, `'gemini'`, `'openrouter'` |
+| `provider` | `string` | `'openai'`, `'openai-responses'`, `'anthropic'`, `'gemini'`, `'openrouter'` |
 | `variables` | `Record<string, string>` | Template variables |
 | `onContextOverflow` | `(info) => string` | Optional callback to transform oversized context values before rendering |
 | `environment` | `string` | Environment override name |
@@ -509,6 +590,7 @@ Renders a prompt for a specific provider. Returns `{ resolved, request?, returnM
 | `history` | `Array<{ role, content }>` | Conversation history |
 | `toolRegistry` | `Record<string, unknown>` | Tool definitions for resolving string tool references |
 | `strict` | `boolean` | Fail on missing variables |
+| `openaiResponses` | `object` | Optional Responses API extras (`previous_response_id`, `conversation`, `instructions`, `parallel_tool_calls`, `max_tool_calls`, `store`, `metadata`, `include`, `background`) |
 
 ### `kit.loadPrompt(path)` / `kit.resolvePrompt(path, options)` / `kit.validatePrompt(path)`
 
@@ -528,13 +610,16 @@ Prompt files use YAML front matter with these fields:
 |-------|------|-------------|
 | `id` | `string` | Unique prompt identifier (required) |
 | `schema_version` | `number` | Schema version, currently `1` |
-| `provider` | `string` | `openai`, `anthropic`, `gemini` (or `google`), `openrouter`, `any` |
+| `provider` | `string` | `openai`, `openai-responses`, `anthropic`, `gemini` (or `google`), `openrouter`, `any` |
 | `model` | `string` | Model name |
 | `fallback_models` | `string[]` | Fallback model list |
 | `reasoning` | `object` | `{ effort, budget_tokens }` |
 | `sampling` | `object` | `{ temperature, top_p, frequency_penalty, presence_penalty, stop, max_output_tokens }` |
-| `response` | `object` | `{ format, stream }` |
+| `response` | `object` | `{ format, stream, schema, schema_name, schema_description, schema_strict }` |
+| `cache` | `object` | Provider-specific cache controls (`openai`, `anthropic`, `gemini`/`google`) |
 | `tools` | `array` | Tool references (string names or inline definitions) |
+| `provider_options` | `object` | Provider-specific non-portable options (`anthropic`, `gemini`, `openrouter`) |
+| `raw` | `object` | Provider-scoped request-body passthrough (`openai`, `openai-responses`, `anthropic`, `gemini`/`google`, `openrouter`) |
 | `mcp` | `object` | MCP server references |
 | `context` | `object` | `{ inputs, history }` — declare expected variables, with optional per-input `max_size`, `trim`, structured or literal `allow_regex`/`deny_regex`, and built-in `non_empty` / `reject_secrets` validators |
 | `includes` | `string[]` | Paths to included prompt files |
@@ -542,6 +627,8 @@ Prompt files use YAML front matter with these fields:
 | `tiers` | `object` | Named tier overrides |
 | `metadata` | `object` | `{ owner, tags, review_required, stable }` |
 
+For `allow_regex` and `deny_regex`, prefer unquoted `/pattern/i` literal form so regex escapes such as `\s` and `\b` stay copyable from tools like regex101. If you use structured `pattern:` form, use single-quoted YAML strings or double each backslash in double-quoted strings.
+
 ## Website
 
 The `website/` directory contains a standalone marketing website for PromptOpsKit.

package/SKILL.md

@@ -8,7 +8,7 @@ description: Guidance for creating and editing promptopskit prompt files, defaul
 This project uses **promptopskit** to manage LLM prompts as code.
 Prompts live in markdown files with YAML front matter, are validated against
 a schema, and render into provider-specific request bodies (OpenAI, Anthropic,
-Gemini, OpenRouter). Follow these instructions when creating or editing prompts.
+Gemini, OpenRouter, and OpenAI Responses). Follow these instructions when creating or editing prompts.
 
 ---
 
@@ -58,13 +58,16 @@ the fields required by that specific file:
 | `id` | string | **yes** | Unique identifier for the prompt |
 | `schema_version` | number | yes | Always `1` |
 | `description` | string | no | Human-readable description |
-| `provider` | enum | no | `openai`, `anthropic`, `google`, `gemini`, `openrouter`, or `any` |
+| `provider` | enum | no | `openai`, `openai-responses`, `anthropic`, `google`, `gemini`, `openrouter`, or `any` |
 | `model` | string | no | Model identifier (e.g. `gpt-5.4`, `claude-sonnet-4-20250514`) |
 | `fallback_models` | string[] | no | Ordered fallback model list |
 | `reasoning` | object | no | `{ effort: low|medium|high, budget_tokens: number }` |
 | `sampling` | object | no | `{ temperature, top_p, frequency_penalty, presence_penalty, stop, max_output_tokens }` |
-| `response` | object | no | `{ format: text|json|markdown, stream: boolean }` |
+| `response` | object | no | `{ format: text|json|markdown, stream: boolean, schema?: object, schema_name?: string, schema_description?: string, schema_strict?: boolean }` |
+| `cache` | object | no | Provider-specific cache controls (`openai`, `anthropic`, `gemini`/`google`) |
 | `tools` | array | no | Tool names (strings) or inline definitions with `{ name, description, input_schema }` |
+| `provider_options` | object | no | Provider-specific advanced options (`anthropic`, `gemini`, `openrouter`) |
+| `raw` | object | no | Provider-scoped request-body passthrough for unmodeled vendor fields |
 | `mcp` | object | no | `{ servers: [string | { name, config }] }` |
 | `context.inputs` | `Array<string | { name, max_size?, trim?, allow_regex?, deny_regex?, non_empty?, reject_secrets? }>` | no | Declared variable names used in templates, with optional size budgets and runtime hardening controls |
 | `context.history` | object | no | `{ max_items: number }` |
@@ -102,7 +105,9 @@ Rules:
 - Use object-form inputs with `max_size` when a variable is likely to grow large and should trigger early warnings
 - Use `trim` to enforce byte budgets before interpolation when `max_size` is set
 - Use `allow_regex` for allowlist checks and `deny_regex` for blocklist checks on risky inputs
-- Prefer structured regexes like `{ pattern, flags }`; `/pattern/i` strings are also accepted and normalized internally
+- Prefer unquoted `/pattern/i` literals for regex validators so backslash escapes such as `\s` and `\b` stay copyable from regex tools
+- Use structured regexes like `{ pattern, flags, return_message }` when the validator needs a fallback message or separate flags
+- In structured `pattern:` YAML, use single quotes for patterns with backslashes or double each backslash in double-quoted strings
 - Use `non_empty: true` for required user text and `reject_secrets: true` for common secret redaction checks
 - When the caller should receive a structured fallback message instead of an exception, use object form with `return_message` on `allow_regex`, `deny_regex`, `non_empty`, or `reject_secrets`
 - Escape literal braces with `\{{` and `\}}`
@@ -124,7 +129,7 @@ At render time, callers can also pass `onContextOverflow` to transform oversized
 
 If a validator declares `return_message`, `renderPrompt()` returns that message in a structured result and omits the provider request instead of throwing for that validation failure. Invalid regex definitions still fail during `validate` and `compile` as `POK013` prompt-authoring errors.
 
-Malformed `allow_regex` and `deny_regex` values fail during `validate` and `compile`, not just at render time. When regex compilation fails, the error includes the prompt id, variable name, field name, and raw configured value.
+Malformed `allow_regex` and `deny_regex` values fail during `validate` and `compile`, not just at render time. When regex compilation fails, the error includes the prompt id, variable name, field name, and raw configured value. Double-quoted YAML regex strings with raw backslashes fail as `POK013`; use `/pattern/i`, single-quoted `pattern: '...'`, or doubled backslashes.
 
 Example: this is the minimal valid shape for a prompt that references
 `{{ pull_request }}` even when provider/model are inherited from defaults:
@@ -182,6 +187,7 @@ prompts/
 Supported default fields:
 - `provider` (front matter) — default provider for the folder
 - `model` (front matter) — default model for the folder
+- `cache` (front matter) — default provider-specific cache hints
 - `metadata` (front matter) — merged with prompt-local metadata
 - `# System instructions` (body section) — used when the prompt has none
 
@@ -229,12 +235,69 @@ tiers:
 ```
 
 Overridable fields: `model`, `fallback_models`, `reasoning`, `sampling`,
-`response`, `tools`.
+`response`, `cache`, `raw`, `tools`, `provider_options`.
 
 Override application order: **base → environment → tier → runtime**.
 
 ---
 
+## Provider-specific fields
+
+Prefer portable fields first:
+
+- Use `sampling` for common sampling controls
+- Use `response.schema`, `response.schema_name`, `response.schema_description`, and `response.schema_strict` for structured output when possible
+- Use `cache` for provider cache hints
+- Use `tools` for tool definitions
+
+Treat `response.schema` as the provider-neutral JSON Schema contract. The adapters emit it through provider-specific request fields: OpenAI/OpenRouter `response_format`, OpenAI Responses `text.format`, Anthropic `output_config.format`, and Gemini `generationConfig.responseJsonSchema`.
+
+Use `provider_options` for known non-portable mappings:
+
+```yaml
+provider_options:
+  anthropic:
+    top_k: 40
+    tool_choice:
+      type: auto
+    output_config:
+      format:
+        type: json_schema
+        schema:
+          type: object
+  gemini:
+    # Use only when Gemini's native schema dialect is required.
+    response_schema:
+      type: object
+    response_json_schema:
+      type: object
+  openrouter:
+    provider:
+      order: ["anthropic", "openai"]
+    transforms: ["middle-out"]
+```
+
+Use `raw` only when a vendor request-body field is important and PromptOpsKit does not model it yet:
+
+```yaml
+raw:
+  openai:
+    service_tier: flex
+  anthropic:
+    service_tier: auto
+  gemini:
+    safetySettings:
+      - category: HARM_CATEGORY_DANGEROUS_CONTENT
+        threshold: BLOCK_ONLY_HIGH
+  openrouter:
+    usage:
+      include: true
+```
+
+Raw blocks are provider-scoped (`openai`, `openai-responses`/`openai_responses`, `anthropic`, `gemini`/`google`, `openrouter`) and are shallow-merged into the final request body after normalized fields. When adding `raw`, include a short note in `# Notes` explaining why a first-class field is not being used.
+
+---
+
 ## Test sidecars
 
 Create a `.test.yaml` file alongside a prompt to define test cases:

package/dist/chunk-F6UJ73Y6.js

@@ -0,0 +1,113 @@
+import {
+  applyRawProviderBody,
+  renderSections,
+  resolveAssetForProvider,
+  withPromptInputSupport
+} from "./chunk-XNNDGDPI.js";
+
+// src/providers/openai-responses.ts
+var openaiResponsesAdapter = withPromptInputSupport({
+  name: "openai-responses",
+  validate(asset, runtime) {
+    const resolvedAsset = resolveAssetForProvider(asset, runtime);
+    const errors = [];
+    const warnings = [];
+    if (!resolvedAsset.model) {
+      errors.push("OpenAI Responses adapter requires a model to be specified.");
+    }
+    if (resolvedAsset.reasoning?.budget_tokens !== void 0) {
+      warnings.push("OpenAI Responses uses reasoning.effort, not budget_tokens. budget_tokens will be ignored.");
+    }
+    if (resolvedAsset.response?.schema !== void 0 && resolvedAsset.response?.format !== "json") {
+      warnings.push("OpenAI Responses response.schema requires response.format: json. schema will still be applied as JSON schema output.");
+    }
+    if (runtime?.openaiResponses?.conversation !== void 0 && runtime?.openaiResponses?.previous_response_id !== void 0) {
+      errors.push('OpenAI Responses options "conversation" and "previous_response_id" cannot both be set.');
+    }
+    return { valid: errors.length === 0, errors, warnings };
+  },
+  render(asset, runtime) {
+    const resolvedAsset = resolveAssetForProvider(asset, runtime);
+    const sections = renderSections(resolvedAsset, {
+      variables: runtime.variables,
+      strict: runtime.strict
+    });
+    const responseOptions = runtime.openaiResponses;
+    const input = [];
+    if (runtime.history) {
+      for (const msg of runtime.history) {
+        input.push({ role: msg.role, content: msg.content });
+      }
+    }
+    if (sections.prompt_template) {
+      input.push({ role: "user", content: sections.prompt_template });
+    }
+    const body = {
+      model: resolvedAsset.model,
+      input
+    };
+    if (responseOptions?.instructions !== void 0) {
+      body.instructions = responseOptions.instructions;
+    } else if (sections.system_instructions) {
+      body.instructions = sections.system_instructions;
+    }
+    if (resolvedAsset.sampling?.temperature !== void 0) body.temperature = resolvedAsset.sampling.temperature;
+    if (resolvedAsset.sampling?.top_p !== void 0) body.top_p = resolvedAsset.sampling.top_p;
+    if (resolvedAsset.sampling?.frequency_penalty !== void 0) body.frequency_penalty = resolvedAsset.sampling.frequency_penalty;
+    if (resolvedAsset.sampling?.presence_penalty !== void 0) body.presence_penalty = resolvedAsset.sampling.presence_penalty;
+    if (resolvedAsset.sampling?.stop !== void 0) body.stop = resolvedAsset.sampling.stop;
+    if (resolvedAsset.sampling?.max_output_tokens !== void 0) body.max_output_tokens = resolvedAsset.sampling.max_output_tokens;
+    if (resolvedAsset.reasoning?.effort) {
+      body.reasoning = { effort: resolvedAsset.reasoning.effort };
+    }
+    if (resolvedAsset.response?.schema) {
+      body.text = {
+        format: {
+          type: "json_schema",
+          name: resolvedAsset.response.schema_name ?? `${resolvedAsset.id}_response`,
+          ...resolvedAsset.response.schema_description ? { description: resolvedAsset.response.schema_description } : {},
+          schema: resolvedAsset.response.schema,
+          strict: resolvedAsset.response.schema_strict ?? true
+        }
+      };
+    } else if (resolvedAsset.response?.format === "json") {
+      body.text = { format: { type: "json_object" } };
+    }
+    if (resolvedAsset.response?.stream !== void 0) {
+      body.stream = resolvedAsset.response.stream;
+    }
+    if (resolvedAsset.tools && resolvedAsset.tools.length > 0) {
+      body.tools = resolvedAsset.tools.map((tool) => {
+        if (typeof tool === "string") {
+          const def = runtime.toolRegistry?.[tool];
+          if (def) return def;
+          return { type: "function", name: tool };
+        }
+        return {
+          type: "function",
+          name: tool.name,
+          description: tool.description,
+          parameters: tool.input_schema
+        };
+      });
+    }
+    if (responseOptions?.previous_response_id !== void 0) body.previous_response_id = responseOptions.previous_response_id;
+    if (responseOptions?.conversation !== void 0) body.conversation = responseOptions.conversation;
+    if (responseOptions?.parallel_tool_calls !== void 0) body.parallel_tool_calls = responseOptions.parallel_tool_calls;
+    if (responseOptions?.max_tool_calls !== void 0) body.max_tool_calls = responseOptions.max_tool_calls;
+    if (responseOptions?.store !== void 0) body.store = responseOptions.store;
+    if (responseOptions?.metadata !== void 0) body.metadata = responseOptions.metadata;
+    if (responseOptions?.include !== void 0) body.include = responseOptions.include;
+    if (responseOptions?.background !== void 0) body.background = responseOptions.background;
+    return {
+      body: applyRawProviderBody(body, resolvedAsset, "openai-responses"),
+      provider: "openai-responses",
+      model: resolvedAsset.model ?? "unknown"
+    };
+  }
+});
+
+export {
+  openaiResponsesAdapter
+};
+//# sourceMappingURL=chunk-F6UJ73Y6.js.map

package/dist/chunk-F6UJ73Y6.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/providers/openai-responses.ts"],"sourcesContent":["import type { ResolvedPromptAsset } from '../schema/index.js';\nimport type {\n  ProviderAdapter,\n  ProviderRequest,\n  ValidationResult,\n  RuntimeRenderOptions,\n} from './types.js';\nimport { renderSections } from '../renderer/index.js';\nimport { resolveAssetForProvider } from './resolve-asset.js';\nimport { withPromptInputSupport } from './prompt-input.js';\nimport { applyRawProviderBody } from './raw.js';\n\n/**\n * OpenAI Responses provider adapter.\n * Produces request bodies compatible with the OpenAI Responses API.\n */\nexport const openaiResponsesAdapter: ProviderAdapter = withPromptInputSupport({\n  name: 'openai-responses',\n\n  validate(asset: ResolvedPromptAsset, runtime?: RuntimeRenderOptions): ValidationResult {\n    const resolvedAsset = resolveAssetForProvider(asset, runtime);\n    const errors: string[] = [];\n    const warnings: string[] = [];\n\n    if (!resolvedAsset.model) {\n      errors.push('OpenAI Responses adapter requires a model to be specified.');\n    }\n\n    if (resolvedAsset.reasoning?.budget_tokens !== undefined) {\n      warnings.push('OpenAI Responses uses reasoning.effort, not budget_tokens. budget_tokens will be ignored.');\n    }\n\n    if (resolvedAsset.response?.schema !== undefined && resolvedAsset.response?.format !== 'json') {\n      warnings.push('OpenAI Responses response.schema requires response.format: json. schema will still be applied as JSON schema output.');\n    }\n\n    if (runtime?.openaiResponses?.conversation !== undefined && runtime?.openaiResponses?.previous_response_id !== undefined) {\n      errors.push('OpenAI Responses options \"conversation\" and \"previous_response_id\" cannot both be set.');\n    }\n\n    return { valid: errors.length === 0, errors, warnings };\n  },\n\n  render(asset: ResolvedPromptAsset, runtime: RuntimeRenderOptions): ProviderRequest {\n    const resolvedAsset = resolveAssetForProvider(asset, runtime);\n    const sections = renderSections(resolvedAsset, {\n      variables: runtime.variables,\n      strict: runtime.strict,\n    });\n    const responseOptions = runtime.openaiResponses;\n\n    const input: Array<Record<string, unknown>> = [];\n\n    if (runtime.history) {\n      for (const msg of runtime.history) {\n        input.push({ role: msg.role, content: msg.content });\n      }\n    }\n\n    if (sections.prompt_template) {\n      input.push({ role: 'user', content: sections.prompt_template });\n    }\n\n    const body: Record<string, unknown> = {\n      model: resolvedAsset.model,\n      input,\n    };\n\n    if (responseOptions?.instructions !== undefined) {\n      body.instructions = responseOptions.instructions;\n    } else if (sections.system_instructions) {\n      body.instructions = sections.system_instructions;\n    }\n\n    // Sampling params\n    if (resolvedAsset.sampling?.temperature !== undefined) body.temperature = resolvedAsset.sampling.temperature;\n    if (resolvedAsset.sampling?.top_p !== undefined) body.top_p = resolvedAsset.sampling.top_p;\n    if (resolvedAsset.sampling?.frequency_penalty !== undefined) body.frequency_penalty = resolvedAsset.sampling.frequency_penalty;\n    if (resolvedAsset.sampling?.presence_penalty !== undefined) body.presence_penalty = resolvedAsset.sampling.presence_penalty;\n    if (resolvedAsset.sampling?.stop !== undefined) body.stop = resolvedAsset.sampling.stop;\n    if (resolvedAsset.sampling?.max_output_tokens !== undefined) body.max_output_tokens = resolvedAsset.sampling.max_output_tokens;\n\n    // Reasoning\n    if (resolvedAsset.reasoning?.effort) {\n      body.reasoning = { effort: resolvedAsset.reasoning.effort };\n    }\n\n    // Structured output / response format\n    if (resolvedAsset.response?.schema) {\n      body.text = {\n        format: {\n          type: 'json_schema',\n          name: resolvedAsset.response.schema_name ?? `${resolvedAsset.id}_response`,\n          ...(resolvedAsset.response.schema_description ? { description: resolvedAsset.response.schema_description } : {}),\n          schema: resolvedAsset.response.schema,\n          strict: resolvedAsset.response.schema_strict ?? true,\n        },\n      };\n    } else if (resolvedAsset.response?.format === 'json') {\n      body.text = { format: { type: 'json_object' } };\n    }\n\n    // Streaming\n    if (resolvedAsset.response?.stream !== undefined) {\n      body.stream = resolvedAsset.response.stream;\n    }\n\n    // Tools\n    if (resolvedAsset.tools && resolvedAsset.tools.length > 0) {\n      body.tools = resolvedAsset.tools.map((tool) => {\n        if (typeof tool === 'string') {\n          const def = runtime.toolRegistry?.[tool];\n          if (def) return def;\n          return { type: 'function', name: tool };\n        }\n        return {\n          type: 'function',\n          name: tool.name,\n          description: tool.description,\n          parameters: tool.input_schema,\n        };\n      });\n    }\n\n    if (responseOptions?.previous_response_id !== undefined) body.previous_response_id = responseOptions.previous_response_id;\n    if (responseOptions?.conversation !== undefined) body.conversation = responseOptions.conversation;\n    if (responseOptions?.parallel_tool_calls !== undefined) body.parallel_tool_calls = responseOptions.parallel_tool_calls;\n    if (responseOptions?.max_tool_calls !== undefined) body.max_tool_calls = responseOptions.max_tool_calls;\n    if (responseOptions?.store !== undefined) body.store = responseOptions.store;\n    if (responseOptions?.metadata !== undefined) body.metadata = responseOptions.metadata;\n    if (responseOptions?.include !== undefined) body.include = responseOptions.include;\n    if (responseOptions?.background !== undefined) body.background = responseOptions.background;\n\n    return {\n      body: applyRawProviderBody(body, resolvedAsset, 'openai-responses'),\n      provider: 'openai-responses',\n      model: resolvedAsset.model ?? 'unknown',\n    };\n  },\n});\n"],"mappings":";;;;;;;;AAgBO,IAAM,yBAA0C,uBAAuB;AAAA,EAC5E,MAAM;AAAA,EAEN,SAAS,OAA4B,SAAkD;AACrF,UAAM,gBAAgB,wBAAwB,OAAO,OAAO;AAC5D,UAAM,SAAmB,CAAC;AAC1B,UAAM,WAAqB,CAAC;AAE5B,QAAI,CAAC,cAAc,OAAO;AACxB,aAAO,KAAK,4DAA4D;AAAA,IAC1E;AAEA,QAAI,cAAc,WAAW,kBAAkB,QAAW;AACxD,eAAS,KAAK,2FAA2F;AAAA,IAC3G;AAEA,QAAI,cAAc,UAAU,WAAW,UAAa,cAAc,UAAU,WAAW,QAAQ;AAC7F,eAAS,KAAK,sHAAsH;AAAA,IACtI;AAEA,QAAI,SAAS,iBAAiB,iBAAiB,UAAa,SAAS,iBAAiB,yBAAyB,QAAW;AACxH,aAAO,KAAK,wFAAwF;AAAA,IACtG;AAEA,WAAO,EAAE,OAAO,OAAO,WAAW,GAAG,QAAQ,SAAS;AAAA,EACxD;AAAA,EAEA,OAAO,OAA4B,SAAgD;AACjF,UAAM,gBAAgB,wBAAwB,OAAO,OAAO;AAC5D,UAAM,WAAW,eAAe,eAAe;AAAA,MAC7C,WAAW,QAAQ;AAAA,MACnB,QAAQ,QAAQ;AAAA,IAClB,CAAC;AACD,UAAM,kBAAkB,QAAQ;AAEhC,UAAM,QAAwC,CAAC;AAE/C,QAAI,QAAQ,SAAS;AACnB,iBAAW,OAAO,QAAQ,SAAS;AACjC,cAAM,KAAK,EAAE,MAAM,IAAI,MAAM,SAAS,IAAI,QAAQ,CAAC;AAAA,MACrD;AAAA,IACF;AAEA,QAAI,SAAS,iBAAiB;AAC5B,YAAM,KAAK,EAAE,MAAM,QAAQ,SAAS,SAAS,gBAAgB,CAAC;AAAA,IAChE;AAEA,UAAM,OAAgC;AAAA,MACpC,OAAO,cAAc;AAAA,MACrB;AAAA,IACF;AAEA,QAAI,iBAAiB,iBAAiB,QAAW;AAC/C,WAAK,eAAe,gBAAgB;AAAA,IACtC,WAAW,SAAS,qBAAqB;AACvC,WAAK,eAAe,SAAS;AAAA,IAC/B;AAGA,QAAI,cAAc,UAAU,gBAAgB,OAAW,MAAK,cAAc,cAAc,SAAS;AACjG,QAAI,cAAc,UAAU,UAAU,OAAW,MAAK,QAAQ,cAAc,SAAS;AACrF,QAAI,cAAc,UAAU,sBAAsB,OAAW,MAAK,oBAAoB,cAAc,SAAS;AAC7G,QAAI,cAAc,UAAU,qBAAqB,OAAW,MAAK,mBAAmB,cAAc,SAAS;AAC3G,QAAI,cAAc,UAAU,SAAS,OAAW,MAAK,OAAO,cAAc,SAAS;AACnF,QAAI,cAAc,UAAU,sBAAsB,OAAW,MAAK,oBAAoB,cAAc,SAAS;AAG7G,QAAI,cAAc,WAAW,QAAQ;AACnC,WAAK,YAAY,EAAE,QAAQ,cAAc,UAAU,OAAO;AAAA,IAC5D;AAGA,QAAI,cAAc,UAAU,QAAQ;AAClC,WAAK,OAAO;AAAA,QACV,QAAQ;AAAA,UACN,MAAM;AAAA,UACN,MAAM,cAAc,SAAS,eAAe,GAAG,cAAc,EAAE;AAAA,UAC/D,GAAI,cAAc,SAAS,qBAAqB,EAAE,aAAa,cAAc,SAAS,mBAAmB,IAAI,CAAC;AAAA,UAC9G,QAAQ,cAAc,SAAS;AAAA,UAC/B,QAAQ,cAAc,SAAS,iBAAiB;AAAA,QAClD;AAAA,MACF;AAAA,IACF,WAAW,cAAc,UAAU,WAAW,QAAQ;AACpD,WAAK,OAAO,EAAE,QAAQ,EAAE,MAAM,cAAc,EAAE;AAAA,IAChD;AAGA,QAAI,cAAc,UAAU,WAAW,QAAW;AAChD,WAAK,SAAS,cAAc,SAAS;AAAA,IACvC;AAGA,QAAI,cAAc,SAAS,cAAc,MAAM,SAAS,GAAG;AACzD,WAAK,QAAQ,cAAc,MAAM,IAAI,CAAC,SAAS;AAC7C,YAAI,OAAO,SAAS,UAAU;AAC5B,gBAAM,MAAM,QAAQ,eAAe,IAAI;AACvC,cAAI,IAAK,QAAO;AAChB,iBAAO,EAAE,MAAM,YAAY,MAAM,KAAK;AAAA,QACxC;AACA,eAAO;AAAA,UACL,MAAM;AAAA,UACN,MAAM,KAAK;AAAA,UACX,aAAa,KAAK;AAAA,UAClB,YAAY,KAAK;AAAA,QACnB;AAAA,MACF,CAAC;AAAA,IACH;AAEA,QAAI,iBAAiB,yBAAyB,OAAW,MAAK,uBAAuB,gBAAgB;AACrG,QAAI,iBAAiB,iBAAiB,OAAW,MAAK,eAAe,gBAAgB;AACrF,QAAI,iBAAiB,wBAAwB,OAAW,MAAK,sBAAsB,gBAAgB;AACnG,QAAI,iBAAiB,mBAAmB,OAAW,MAAK,iBAAiB,gBAAgB;AACzF,QAAI,iBAAiB,UAAU,OAAW,MAAK,QAAQ,gBAAgB;AACvE,QAAI,iBAAiB,aAAa,OAAW,MAAK,WAAW,gBAAgB;AAC7E,QAAI,iBAAiB,YAAY,OAAW,MAAK,UAAU,gBAAgB;AAC3E,QAAI,iBAAiB,eAAe,OAAW,MAAK,aAAa,gBAAgB;AAEjF,WAAO;AAAA,MACL,MAAM,qBAAqB,MAAM,eAAe,kBAAkB;AAAA,MAClE,UAAU;AAAA,MACV,OAAO,cAAc,SAAS;AAAA,IAChC;AAAA,EACF;AACF,CAAC;","names":[]}

package/dist/chunk-HLNVKKMS.js

@@ -0,0 +1,145 @@
+import {
+  applyRawProviderBody,
+  renderSections,
+  resolveAssetForProvider,
+  withPromptInputSupport
+} from "./chunk-XNNDGDPI.js";
+
+// src/providers/anthropic.ts
+var anthropicAdapter = withPromptInputSupport({
+  name: "anthropic",
+  validate(asset, runtime) {
+    const resolvedAsset = resolveAssetForProvider(asset, runtime);
+    const errors = [];
+    const warnings = [];
+    if (!resolvedAsset.model) {
+      errors.push("Anthropic adapter requires a model to be specified.");
+    }
+    if (resolvedAsset.sampling?.frequency_penalty !== void 0) {
+      warnings.push("Anthropic does not support frequency_penalty. It will be ignored.");
+    }
+    if (resolvedAsset.sampling?.presence_penalty !== void 0) {
+      warnings.push("Anthropic does not support presence_penalty. It will be ignored.");
+    }
+    if (resolvedAsset.reasoning?.effort !== void 0) {
+      warnings.push("Anthropic uses budget_tokens for thinking, not effort. effort will be mapped approximately.");
+    }
+    if (resolvedAsset.response?.schema !== void 0 && resolvedAsset.response?.format !== "json") {
+      warnings.push("Anthropic response.schema is mapped to output_config.format and should usually be paired with response.format: json.");
+    }
+    if (resolvedAsset.provider_options?.anthropic?.top_k !== void 0 && resolvedAsset.provider_options.anthropic.top_k < 0) {
+      errors.push("Anthropic provider_options.top_k must be >= 0.");
+    }
+    return { valid: errors.length === 0, errors, warnings };
+  },
+  render(asset, runtime) {
+    const resolvedAsset = resolveAssetForProvider(asset, runtime);
+    const sections = renderSections(resolvedAsset, {
+      variables: runtime.variables,
+      strict: runtime.strict
+    });
+    const messages = [];
+    const anthropicCacheConfig = resolvedAsset.cache?.anthropic;
+    const cacheType = anthropicCacheConfig?.type ?? "ephemeral";
+    const cacheControl = anthropicCacheConfig ? {
+      type: cacheType,
+      ...anthropicCacheConfig.ttl ? { ttl: anthropicCacheConfig.ttl } : {}
+    } : void 0;
+    const cacheMode = anthropicCacheConfig?.mode ?? "automatic";
+    if (runtime.history) {
+      for (const msg of runtime.history) {
+        messages.push({ role: msg.role, content: msg.content });
+      }
+    }
+    if (sections.prompt_template) {
+      if (cacheControl && cacheMode === "explicit" && anthropicCacheConfig?.cache_prompt_template) {
+        messages.push({
+          role: "user",
+          content: [{ type: "text", text: sections.prompt_template, cache_control: cacheControl }]
+        });
+      } else {
+        messages.push({ role: "user", content: sections.prompt_template });
+      }
+    }
+    const body = {
+      model: resolvedAsset.model,
+      messages
+    };
+    if (sections.system_instructions) {
+      if (cacheControl && cacheMode === "explicit" && anthropicCacheConfig?.cache_system_instructions !== false) {
+        body.system = [{ type: "text", text: sections.system_instructions, cache_control: cacheControl }];
+      } else {
+        body.system = sections.system_instructions;
+      }
+    }
+    if (resolvedAsset.sampling?.temperature !== void 0) body.temperature = resolvedAsset.sampling.temperature;
+    if (resolvedAsset.sampling?.top_p !== void 0) body.top_p = resolvedAsset.sampling.top_p;
+    if (resolvedAsset.sampling?.stop !== void 0) body.stop_sequences = resolvedAsset.sampling.stop;
+    if (resolvedAsset.sampling?.max_output_tokens !== void 0) {
+      body.max_tokens = resolvedAsset.sampling.max_output_tokens;
+    } else {
+      body.max_tokens = 4096;
+    }
+    if (resolvedAsset.reasoning?.budget_tokens) {
+      body.thinking = {
+        type: "enabled",
+        budget_tokens: resolvedAsset.reasoning.budget_tokens
+      };
+    }
+    if (resolvedAsset.provider_options?.anthropic?.top_k !== void 0) {
+      body.top_k = resolvedAsset.provider_options.anthropic.top_k;
+    }
+    if (resolvedAsset.provider_options?.anthropic?.output_config !== void 0) {
+      body.output_config = resolvedAsset.provider_options.anthropic.output_config;
+    } else if (resolvedAsset.response?.schema !== void 0) {
+      body.output_config = {
+        format: {
+          type: "json_schema",
+          schema: resolvedAsset.response.schema
+        }
+      };
+    }
+    if (resolvedAsset.response?.stream !== void 0) {
+      body.stream = resolvedAsset.response.stream;
+    }
+    if (cacheControl && cacheMode === "automatic") {
+      body.cache_control = cacheControl;
+    }
+    if (resolvedAsset.tools && resolvedAsset.tools.length > 0) {
+      body.tools = resolvedAsset.tools.map((tool) => {
+        if (typeof tool === "string") {
+          const def = runtime.toolRegistry?.[tool];
+          if (def) {
+            if (cacheControl && cacheMode === "explicit" && anthropicCacheConfig?.cache_tools) {
+              return { ...def, cache_control: cacheControl };
+            }
+            return def;
+          }
+          return {
+            name: tool,
+            ...cacheControl && cacheMode === "explicit" && anthropicCacheConfig?.cache_tools ? { cache_control: cacheControl } : {}
+          };
+        }
+        return {
+          name: tool.name,
+          description: tool.description,
+          input_schema: tool.input_schema ?? { type: "object", properties: {} },
+          ...cacheControl && cacheMode === "explicit" && anthropicCacheConfig?.cache_tools ? { cache_control: cacheControl } : {}
+        };
+      });
+    }
+    if (resolvedAsset.provider_options?.anthropic?.tool_choice !== void 0) {
+      body.tool_choice = resolvedAsset.provider_options.anthropic.tool_choice;
+    }
+    return {
+      body: applyRawProviderBody(body, resolvedAsset, "anthropic"),
+      provider: "anthropic",
+      model: resolvedAsset.model ?? "unknown"
+    };
+  }
+});
+
+export {
+  anthropicAdapter
+};
+//# sourceMappingURL=chunk-HLNVKKMS.js.map

package/dist/chunk-HLNVKKMS.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/providers/anthropic.ts"],"sourcesContent":["import type { ResolvedPromptAsset } from '../schema/index.js';\nimport type {\n  ProviderAdapter,\n  ProviderRequest,\n  ValidationResult,\n  RuntimeRenderOptions,\n} from './types.js';\nimport { renderSections } from '../renderer/index.js';\nimport { resolveAssetForProvider } from './resolve-asset.js';\nimport { withPromptInputSupport } from './prompt-input.js';\nimport { applyRawProviderBody } from './raw.js';\n\n/**\n * Anthropic provider adapter.\n * Produces request bodies compatible with the Anthropic Messages API.\n */\nexport const anthropicAdapter: ProviderAdapter = withPromptInputSupport({\n  name: 'anthropic',\n\n  validate(asset: ResolvedPromptAsset, runtime?: RuntimeRenderOptions): ValidationResult {\n    const resolvedAsset = resolveAssetForProvider(asset, runtime);\n    const errors: string[] = [];\n    const warnings: string[] = [];\n\n    if (!resolvedAsset.model) {\n      errors.push('Anthropic adapter requires a model to be specified.');\n    }\n\n    if (resolvedAsset.sampling?.frequency_penalty !== undefined) {\n      warnings.push('Anthropic does not support frequency_penalty. It will be ignored.');\n    }\n    if (resolvedAsset.sampling?.presence_penalty !== undefined) {\n      warnings.push('Anthropic does not support presence_penalty. It will be ignored.');\n    }\n    if (resolvedAsset.reasoning?.effort !== undefined) {\n      warnings.push('Anthropic uses budget_tokens for thinking, not effort. effort will be mapped approximately.');\n    }\n    if (resolvedAsset.response?.schema !== undefined && resolvedAsset.response?.format !== 'json') {\n      warnings.push('Anthropic response.schema is mapped to output_config.format and should usually be paired with response.format: json.');\n    }\n\n    if (resolvedAsset.provider_options?.anthropic?.top_k !== undefined && resolvedAsset.provider_options.anthropic.top_k < 0) {\n      errors.push('Anthropic provider_options.top_k must be >= 0.');\n    }\n\n    return { valid: errors.length === 0, errors, warnings };\n  },\n\n  render(asset: ResolvedPromptAsset, runtime: RuntimeRenderOptions): ProviderRequest {\n    const resolvedAsset = resolveAssetForProvider(asset, runtime);\n    const sections = renderSections(resolvedAsset, {\n      variables: runtime.variables,\n      strict: runtime.strict,\n    });\n\n    const messages: Array<Record<string, unknown>> = [];\n    const anthropicCacheConfig = resolvedAsset.cache?.anthropic;\n    const cacheType = anthropicCacheConfig?.type ?? 'ephemeral';\n    const cacheControl = anthropicCacheConfig\n      ? {\n        type: cacheType,\n        ...(anthropicCacheConfig.ttl ? { ttl: anthropicCacheConfig.ttl } : {}),\n      }\n      : undefined;\n    const cacheMode = anthropicCacheConfig?.mode ?? 'automatic';\n\n    // History\n    if (runtime.history) {\n      for (const msg of runtime.history) {\n        messages.push({ role: msg.role, content: msg.content });\n      }\n    }\n\n    // User message (prompt template)\n    if (sections.prompt_template) {\n      if (cacheControl && cacheMode === 'explicit' && anthropicCacheConfig?.cache_prompt_template) {\n        messages.push({\n          role: 'user',\n          content: [{ type: 'text', text: sections.prompt_template, cache_control: cacheControl }],\n        });\n      } else {\n        messages.push({ role: 'user', content: sections.prompt_template });\n      }\n    }\n\n    const body: Record<string, unknown> = {\n      model: resolvedAsset.model,\n      messages,\n    };\n\n    // System goes as top-level field in Anthropic\n    if (sections.system_instructions) {\n      if (cacheControl && cacheMode === 'explicit' && anthropicCacheConfig?.cache_system_instructions !== false) {\n        body.system = [{ type: 'text', text: sections.system_instructions, cache_control: cacheControl }];\n      } else {\n        body.system = sections.system_instructions;\n      }\n    }\n\n    // Sampling params\n    if (resolvedAsset.sampling?.temperature !== undefined) body.temperature = resolvedAsset.sampling.temperature;\n    if (resolvedAsset.sampling?.top_p !== undefined) body.top_p = resolvedAsset.sampling.top_p;\n    if (resolvedAsset.sampling?.stop !== undefined) body.stop_sequences = resolvedAsset.sampling.stop;\n    if (resolvedAsset.sampling?.max_output_tokens !== undefined) {\n      body.max_tokens = resolvedAsset.sampling.max_output_tokens;\n    } else {\n      // Anthropic requires max_tokens\n      body.max_tokens = 4096;\n    }\n\n    // Thinking/reasoning\n    if (resolvedAsset.reasoning?.budget_tokens) {\n      body.thinking = {\n        type: 'enabled',\n        budget_tokens: resolvedAsset.reasoning.budget_tokens,\n      };\n    }\n\n    // Provider-specific options\n    if (resolvedAsset.provider_options?.anthropic?.top_k !== undefined) {\n      body.top_k = resolvedAsset.provider_options.anthropic.top_k;\n    }\n\n    if (resolvedAsset.provider_options?.anthropic?.output_config !== undefined) {\n      body.output_config = resolvedAsset.provider_options.anthropic.output_config;\n    } else if (resolvedAsset.response?.schema !== undefined) {\n      body.output_config = {\n        format: {\n          type: 'json_schema',\n          schema: resolvedAsset.response.schema,\n        },\n      };\n    }\n\n    // Streaming\n    if (resolvedAsset.response?.stream !== undefined) {\n      body.stream = resolvedAsset.response.stream;\n    }\n\n    if (cacheControl && cacheMode === 'automatic') {\n      body.cache_control = cacheControl;\n    }\n\n    // Tools\n    if (resolvedAsset.tools && resolvedAsset.tools.length > 0) {\n      body.tools = resolvedAsset.tools.map((tool) => {\n        if (typeof tool === 'string') {\n          const def = runtime.toolRegistry?.[tool];\n          if (def) {\n            if (cacheControl && cacheMode === 'explicit' && anthropicCacheConfig?.cache_tools) {\n              return { ...(def as Record<string, unknown>), cache_control: cacheControl };\n            }\n            return def;\n          }\n          return {\n            name: tool,\n            ...(cacheControl && cacheMode === 'explicit' && anthropicCacheConfig?.cache_tools\n              ? { cache_control: cacheControl }\n              : {}),\n          };\n        }\n        return {\n          name: tool.name,\n          description: tool.description,\n          input_schema: tool.input_schema ?? { type: 'object', properties: {} },\n          ...(cacheControl && cacheMode === 'explicit' && anthropicCacheConfig?.cache_tools\n            ? { cache_control: cacheControl }\n            : {}),\n        };\n      });\n    }\n\n    if (resolvedAsset.provider_options?.anthropic?.tool_choice !== undefined) {\n      body.tool_choice = resolvedAsset.provider_options.anthropic.tool_choice;\n    }\n\n    return {\n      body: applyRawProviderBody(body, resolvedAsset, 'anthropic'),\n      provider: 'anthropic',\n      model: resolvedAsset.model ?? 'unknown',\n    };\n  },\n});\n"],"mappings":";;;;;;;;AAgBO,IAAM,mBAAoC,uBAAuB;AAAA,EACtE,MAAM;AAAA,EAEN,SAAS,OAA4B,SAAkD;AACrF,UAAM,gBAAgB,wBAAwB,OAAO,OAAO;AAC5D,UAAM,SAAmB,CAAC;AAC1B,UAAM,WAAqB,CAAC;AAE5B,QAAI,CAAC,cAAc,OAAO;AACxB,aAAO,KAAK,qDAAqD;AAAA,IACnE;AAEA,QAAI,cAAc,UAAU,sBAAsB,QAAW;AAC3D,eAAS,KAAK,mEAAmE;AAAA,IACnF;AACA,QAAI,cAAc,UAAU,qBAAqB,QAAW;AAC1D,eAAS,KAAK,kEAAkE;AAAA,IAClF;AACA,QAAI,cAAc,WAAW,WAAW,QAAW;AACjD,eAAS,KAAK,6FAA6F;AAAA,IAC7G;AACA,QAAI,cAAc,UAAU,WAAW,UAAa,cAAc,UAAU,WAAW,QAAQ;AAC7F,eAAS,KAAK,sHAAsH;AAAA,IACtI;AAEA,QAAI,cAAc,kBAAkB,WAAW,UAAU,UAAa,cAAc,iBAAiB,UAAU,QAAQ,GAAG;AACxH,aAAO,KAAK,gDAAgD;AAAA,IAC9D;AAEA,WAAO,EAAE,OAAO,OAAO,WAAW,GAAG,QAAQ,SAAS;AAAA,EACxD;AAAA,EAEA,OAAO,OAA4B,SAAgD;AACjF,UAAM,gBAAgB,wBAAwB,OAAO,OAAO;AAC5D,UAAM,WAAW,eAAe,eAAe;AAAA,MAC7C,WAAW,QAAQ;AAAA,MACnB,QAAQ,QAAQ;AAAA,IAClB,CAAC;AAED,UAAM,WAA2C,CAAC;AAClD,UAAM,uBAAuB,cAAc,OAAO;AAClD,UAAM,YAAY,sBAAsB,QAAQ;AAChD,UAAM,eAAe,uBACjB;AAAA,MACA,MAAM;AAAA,MACN,GAAI,qBAAqB,MAAM,EAAE,KAAK,qBAAqB,IAAI,IAAI,CAAC;AAAA,IACtE,IACE;AACJ,UAAM,YAAY,sBAAsB,QAAQ;AAGhD,QAAI,QAAQ,SAAS;AACnB,iBAAW,OAAO,QAAQ,SAAS;AACjC,iBAAS,KAAK,EAAE,MAAM,IAAI,MAAM,SAAS,IAAI,QAAQ,CAAC;AAAA,MACxD;AAAA,IACF;AAGA,QAAI,SAAS,iBAAiB;AAC5B,UAAI,gBAAgB,cAAc,cAAc,sBAAsB,uBAAuB;AAC3F,iBAAS,KAAK;AAAA,UACZ,MAAM;AAAA,UACN,SAAS,CAAC,EAAE,MAAM,QAAQ,MAAM,SAAS,iBAAiB,eAAe,aAAa,CAAC;AAAA,QACzF,CAAC;AAAA,MACH,OAAO;AACL,iBAAS,KAAK,EAAE,MAAM,QAAQ,SAAS,SAAS,gBAAgB,CAAC;AAAA,MACnE;AAAA,IACF;AAEA,UAAM,OAAgC;AAAA,MACpC,OAAO,cAAc;AAAA,MACrB;AAAA,IACF;AAGA,QAAI,SAAS,qBAAqB;AAChC,UAAI,gBAAgB,cAAc,cAAc,sBAAsB,8BAA8B,OAAO;AACzG,aAAK,SAAS,CAAC,EAAE,MAAM,QAAQ,MAAM,SAAS,qBAAqB,eAAe,aAAa,CAAC;AAAA,MAClG,OAAO;AACL,aAAK,SAAS,SAAS;AAAA,MACzB;AAAA,IACF;AAGA,QAAI,cAAc,UAAU,gBAAgB,OAAW,MAAK,cAAc,cAAc,SAAS;AACjG,QAAI,cAAc,UAAU,UAAU,OAAW,MAAK,QAAQ,cAAc,SAAS;AACrF,QAAI,cAAc,UAAU,SAAS,OAAW,MAAK,iBAAiB,cAAc,SAAS;AAC7F,QAAI,cAAc,UAAU,sBAAsB,QAAW;AAC3D,WAAK,aAAa,cAAc,SAAS;AAAA,IAC3C,OAAO;AAEL,WAAK,aAAa;AAAA,IACpB;AAGA,QAAI,cAAc,WAAW,eAAe;AAC1C,WAAK,WAAW;AAAA,QACd,MAAM;AAAA,QACN,eAAe,cAAc,UAAU;AAAA,MACzC;AAAA,IACF;AAGA,QAAI,cAAc,kBAAkB,WAAW,UAAU,QAAW;AAClE,WAAK,QAAQ,cAAc,iBAAiB,UAAU;AAAA,IACxD;AAEA,QAAI,cAAc,kBAAkB,WAAW,kBAAkB,QAAW;AAC1E,WAAK,gBAAgB,cAAc,iBAAiB,UAAU;AAAA,IAChE,WAAW,cAAc,UAAU,WAAW,QAAW;AACvD,WAAK,gBAAgB;AAAA,QACnB,QAAQ;AAAA,UACN,MAAM;AAAA,UACN,QAAQ,cAAc,SAAS;AAAA,QACjC;AAAA,MACF;AAAA,IACF;AAGA,QAAI,cAAc,UAAU,WAAW,QAAW;AAChD,WAAK,SAAS,cAAc,SAAS;AAAA,IACvC;AAEA,QAAI,gBAAgB,cAAc,aAAa;AAC7C,WAAK,gBAAgB;AAAA,IACvB;AAGA,QAAI,cAAc,SAAS,cAAc,MAAM,SAAS,GAAG;AACzD,WAAK,QAAQ,cAAc,MAAM,IAAI,CAAC,SAAS;AAC7C,YAAI,OAAO,SAAS,UAAU;AAC5B,gBAAM,MAAM,QAAQ,eAAe,IAAI;AACvC,cAAI,KAAK;AACP,gBAAI,gBAAgB,cAAc,cAAc,sBAAsB,aAAa;AACjF,qBAAO,EAAE,GAAI,KAAiC,eAAe,aAAa;AAAA,YAC5E;AACA,mBAAO;AAAA,UACT;AACA,iBAAO;AAAA,YACL,MAAM;AAAA,YACN,GAAI,gBAAgB,cAAc,cAAc,sBAAsB,cAClE,EAAE,eAAe,aAAa,IAC9B,CAAC;AAAA,UACP;AAAA,QACF;AACA,eAAO;AAAA,UACL,MAAM,KAAK;AAAA,UACX,aAAa,KAAK;AAAA,UAClB,cAAc,KAAK,gBAAgB,EAAE,MAAM,UAAU,YAAY,CAAC,EAAE;AAAA,UACpE,GAAI,gBAAgB,cAAc,cAAc,sBAAsB,cAClE,EAAE,eAAe,aAAa,IAC9B,CAAC;AAAA,QACP;AAAA,MACF,CAAC;AAAA,IACH;AAEA,QAAI,cAAc,kBAAkB,WAAW,gBAAgB,QAAW;AACxE,WAAK,cAAc,cAAc,iBAAiB,UAAU;AAAA,IAC9D;AAEA,WAAO;AAAA,MACL,MAAM,qBAAqB,MAAM,eAAe,WAAW;AAAA,MAC3D,UAAU;AAAA,MACV,OAAO,cAAc,SAAS;AAAA,IAChC;AAAA,EACF;AACF,CAAC;","names":[]}