### Narev Router Skill - Example Agent Prompts

Source: https://context7.com/narevai/skills/llms.txt

These examples demonstrate how agent prompts are routed by the `narev` skill to the appropriate Narev Cloud documentation paths or specific skills.

```text
# Example agent prompts that invoke the narev router:

"Which Narev skill should I use for billing middleware?"
→ Routes to: /sdk/ai-billing/index

"How do I set up usage-based billing concepts?"
→ Routes to: /platform/concepts/usage-based-billing

"Show me the Polar billing integration."
→ Routes to: /platform/billing/overview

"How do I use the routing API?"
→ Routes to: /platform/routing/introduction

"What is Narev Self-Hosted?"
→ Routes to: /narev-oss/index

# The router also checks package.json for @ai-billing/core and
# @ai-billing/<provider> to select the correct major-version SDK docs.
```

--------------------------------

### Install Agent Skills

Source: https://github.com/narevai/skills/blob/main/README.md

Use this command to add Narev Skills to your Agent Skills environment.

```bash
npx skills add narevai/skills
```

--------------------------------

### Manual Installation for Claude Code

Source: https://github.com/narevai/skills/blob/main/README.md

Clone the Narev Skills repository into your local Claude skills directory for manual integration.

```bash
git clone https://github.com/narevai/skills ~/.claude/skills/narev
```

--------------------------------

### Install Codex Plugin

Source: https://github.com/narevai/skills/blob/main/README.md

Add Narev Skills as a plugin to Codex. After installation, restart Codex and enable the skill from the plugins section.

```bash
codex plugin marketplace add narevai/skills
```

--------------------------------

### Anthropic Plugin Marketplace Format

Source: https://github.com/narevai/skills/blob/main/CLAUDE.md

Example of the `.claude-plugin/marketplace.json` file, which lists Narev skills for the Anthropic plugin system. It shows how to group skills under 'narev' and 'pricing' plugins.

```json
.claude-plugin/marketplace.json — Anthropic plugin format. The **narev** plugin lists the router (`skills/core/narev/`); the **pricing** plugin lists the two pricing skills under `skills/pricing/`.
```

--------------------------------

### Codex Marketplace Registry

Source: https://github.com/narevai/skills/blob/main/AGENTS.md

Details the Codex marketplace registry configuration for installing the Narev skills plugin from a Git URL.

```json
.agents/plugins/marketplace.json — Codex marketplace registry for installing the plugin from the Git URL.
```

--------------------------------

### Get Available Model Pricing Information

Source: https://context7.com/narevai/skills/llms.txt

Retrieves a list of available models and their pricing details, useful for identifying correct model IDs and subproviders.

```APIDOC
## GET /api/models/pricing

### Description
Fetches a list of models and their associated pricing information, including subprovider details for multi-host models.

### Method
GET

### Endpoint
https://www.narev.ai/api/models/pricing

### Parameters
#### Query Parameters
- **model_id** (string) - Optional - Filters the results to a specific model ID.

### Response
#### Success Response (200)
- **data** (array) - A list of model pricing objects.
  - Each object contains details like `modelId`, `provider`, `subprovider`, and pricing information.

### Response Example
```json
[
  {
    "modelId": "llama-3.1-70b",
    "provider": "meta",
    "subprovider": "together",
    "pricing": {
      "input": 0.0000006,
      "output": 0.0000006,
      "request": null,
      "inputCacheRead": 0.0000003,
      "inputCacheWrite": null,
      "internalReasoning": null,
      "webSearch": null
    }
  }
  // ... more models
]
```
```

--------------------------------

### CI Workflow for Scheduled Pricing Refresh

Source: https://context7.com/narevai/skills/llms.txt

Automates the refresh of LLM pricing snapshots every Monday. It checks out code, sets up Node.js, installs dependencies, runs a pricing update script, and opens a pull request if changes are detected.

```yaml
# .github/workflows/update-pricing.yml
name: Refresh LLM Pricing

on:
  schedule:
    - cron: "0 6 * * 1"   # Every Monday at 06:00 UTC
  workflow_dispatch:

jobs:
  update:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-node@v4
        with:
          node-version: 20
      - run: pnpm install
      - run: pnpm tsx scripts/update-pricing.ts
        # Non-zero exit on HTTP error causes the job to fail visibly
      - name: Open PR if pricing changed
        uses: peter-evans/create-pull-request@v6
        with:
          commit-message: "chore: refresh LLM pricing snapshot"
          title: "chore: refresh LLM pricing snapshot"
          body: "Auto-generated by the weekly pricing refresh workflow."
          branch: chore/refresh-pricing

```

--------------------------------

### Calculate Claude Sonnet 4 Cost

Source: https://context7.com/narevai/skills/llms.txt

This example demonstrates calculating the cost for an Anthropic Claude Sonnet 4 model call. Always include all required fields in the usage object.

```bash
curl -X POST 'https://www.narev.ai/api/models/pricing/calculate' \
  -H 'Content-Type: application/json' \
  -d '{
    "modelId": "claude-sonnet-4-5",
    "provider": "anthropic",
    "usage": {
      "promptTokens": 1200,
      "completionTokens": 400,
      "cacheReadTokens": 0,
      "cacheWriteTokens": 0,
      "reasoningTokens": 0
    }
  }'
```

--------------------------------

### List LLM Model Pricing with cURL

Source: https://github.com/narevai/skills/blob/main/skills/pricing/narev-lookup-llm-pricing/SKILL.md

Use this command to fetch a list of LLM models and their pricing details from the Narev API. You can filter results by model ID and provider. Ensure you have cURL installed.

```bash
curl -G 'https://www.narev.ai/api/models/pricing' \
  --data-urlencode 'model_id=gpt-4o' \
  --data-urlencode 'provider=openai'
```

--------------------------------

### Handle Pricing API Errors

Source: https://context7.com/narevai/skills/llms.txt

Examples of common error responses from the Narev Pricing API, including usage validation failures, enterprise-only models, and not found errors. Always provide zero for unused usage fields.

```bash
# 400 Bad Request — usage validation failure
# Cause: missing required integer fields in usage object
# Fix: pass 0 for all unused fields, never null or omit
# Bad body:  { "usage": { "promptTokens": 100, "completionTokens": 50 } }
# Good body: { "usage": { "promptTokens": 100, "completionTokens": 50,
#                         "cacheReadTokens": 0, "cacheWriteTokens": 0,
#                         "reasoningTokens": 0 } }
```

```bash
# 402 Payment Required — enterprise-only model
# Response: { "error": "...", "pricing": null }
# Action: contact Narev for enterprise access; do not invent rates
```

```bash
# 404 Not Found — no public pricing for this modelId + provider combo
# Action: GET /api/models/pricing to verify correct IDs and subprovider;
#         for multi-host models (llama, etc.) subprovider is required
```

```bash
# subprovider resolution example for llama-3.1-70b:
curl -G 'https://www.narev.ai/api/models/pricing' \
  --data-urlencode 'model_id=llama-3.1-70b'
# → inspect data[].subprovider to find "bedrock", "openrouter", "together", etc.
# → then POST calculate with the correct subprovider value
```

--------------------------------

### TypeScript Snapshot Script for Pricing Registry

Source: https://context7.com/narevai/skills/llms.txt

Fetches full catalog pricing data from the narev.ai API and writes it to src/pricing.json. Handles pagination and converts USD per token to per-million-token rates. Use this script to keep pricing information up-to-date.

```typescript
// scripts/update-pricing.ts
// Run: pnpm tsx scripts/update-pricing.ts
// Fetches full catalog and writes pricing.json with per-million-token rates.

interface PricingEntry {
  inputPerMTok: number;
  outputPerMTok: number;
  cacheReadPerMTok?: number;
  cacheWritePerMTok?: number;
  requestFlat?: number;
  discount?: number;
}

async function fetchAllPricing(provider?: string): Promise<Record<string, PricingEntry>> {
  const registry: Record<string, PricingEntry> = {};
  let page = 1;
  const limit = 1000;

  while (true) {
    const params = new URLSearchParams({ page: String(page), limit: String(limit) });
    if (provider) params.set("provider", provider);

    const res = await fetch(`https://www.narev.ai/api/models/pricing?${params}`);
    if (!res.ok) {
      console.error(`HTTP ${res.status}`);
      process.exit(1); // Non-zero exit so CI fails visibly
    }

    const { data, meta } = await res.json();

    for (const row of data) {
      if (!row.pricing) continue; // Skip enterprise-only (null pricing)

      // Key as "provider/model_id" to avoid last-write-wins for shared model IDs
      const key = `${row.provider}/${row.model_id}`;

      // API returns USD per token; multiply by 1_000_000 for per-million units
      registry[key] = {
        inputPerMTok:      (row.pricing.price_prompt      ?? 0) * 1_000_000,
        outputPerMTok:     (row.pricing.price_completion  ?? 0) * 1_000_000,
        cacheReadPerMTok:  row.pricing.price_input_cache_read
                             ? row.pricing.price_input_cache_read * 1_000_000
                             : undefined,
        cacheWritePerMTok: row.pricing.price_input_cache_write
                             ? row.pricing.price_input_cache_write * 1_000_000
                             : undefined,
        requestFlat:       row.pricing.pricing_request ?? undefined,
        discount:          row.pricing.pricing_discount ?? undefined,
      };
    }

    if (page >= meta.total_pages) break;
    page++;
  }

  return registry;
}

const registry = await fetchAllPricing(); // or pass "openai" / "anthropic" for partial refresh
const banner = "// AUTO-GENERATED by scripts/update-pricing.ts — do not edit manually\n";
await Bun.write("src/pricing.json", JSON.stringify(registry, null, 2));
console.log(`Wrote ${Object.keys(registry).length} entries to src/pricing.json`);

```

--------------------------------

### Python Snapshot Script for Pricing Registry

Source: https://context7.com/narevai/skills/llms.txt

Fetches pricing data from the narev.ai API and writes it to src/pricing.json. This script handles pagination and converts USD per token to per-million-token rates. It's useful for maintaining an up-to-date pricing registry.

```python
# scripts/update_pricing.py
# Run: uv run python scripts/update_pricing.py
# Writes src/pricing.json with per-million-token rates.

import json, sys
from pathlib import Path
import httpx

def fetch_all_pricing(provider: str | None = None) -> dict:
    registry = {}
    page, limit = 1, 1000

    while True:
        params = {"page": page, "limit": limit}
        if provider:
            params["provider"] = provider

        r = httpx.get("https://www.narev.ai/api/models/pricing", params=params)
        if r.status_code != 200:
            print(f"HTTP {r.status_code}: {r.text}", file=sys.stderr)
            sys.exit(1)  # Non-zero exit so CI fails visibly

        body = r.json()
        data, meta = body["data"], body["meta"]

        for row in data:
            if not row.get("pricing"):
                continue  # Skip null pricing (enterprise-only)

            # Namespace key to avoid collisions across providers
            key = f"{row['provider']}/{row['model_id']}"
            p = row["pricing"]

            # API is USD per token → multiply by 1_000_000 for per-million units
            registry[key] = {
                "inputPerMTok":  (p.get("price_prompt") or 0) * 1_000_000,
                "outputPerMTok": (p.get("price_completion") or 0) * 1_000_000,
                **({"cacheReadPerMTok":  p["price_input_cache_read"]  * 1_000_000}
                   if p.get("price_input_cache_read") else {}),
                **({"cacheWritePerMTok": p["price_input_cache_write"] * 1_000_000}
                   if p.get("price_input_cache_write") else {}),
                **({"requestFlat": p["pricing_request"]} if p.get("pricing_request") else {}),
                **({"discount":    p["pricing_discount"]} if p.get("pricing_discount") else {}),
            }

        if page >= meta["total_pages"]:
            break
        page += 1

    return registry

registry = fetch_all_pricing()  # Pass "openai" or "anthropic" for scoped refresh
Path("src/pricing.json").write_text(
    "// AUTO-GENERATED — do not edit manually\n" + json.dumps(registry, indent=2)
)
print(f"Wrote {len(registry)} entries to src/pricing.json")

```

--------------------------------

### List LLM Pricing with Filters (Narev API)

Source: https://context7.com/narevai/skills/llms.txt

Use this `curl` command to query the Narev Pricing API for specific models and providers. Authentication is not required. Rates are returned in USD per token.

```bash
# List pricing for gpt-4o on OpenAI
curl -G 'https://www.narev.ai/api/models/pricing' \
  --data-urlencode 'model_id=gpt-4o' \
  --data-urlencode 'provider=openai'
```

```json
# Example response:
# {
#   "data": [
#     {
#       "model_id": "gpt-4o",
#       "provider": "openai",
#       "subprovider": "OpenAI",
#       "pricing": {
#         "price_prompt": 0.0000025,
#         "price_completion": 0.00001,
#         "price_input_cache_read": 0.00000125,
#         "price_input_cache_write": null,
#         "price_internal_reasoning": null,
#         "pricing_request": null,
#         "price_web_search": null,
#         "pricing_discount": 0
#       }
#     }
#   ],
#   "meta": { "page": 1, "limit": 100, "total": 1, "total_pages": 1 }
# }

# Convert to USD per million tokens for display:
# input:  0.0000025  * 1_000_000 = $2.50 / 1M tokens
# output: 0.00001    * 1_000_000 = $10.00 / 1M tokens
```

```bash
# List all providers for a multi-host model (subprovider matters)
curl -G 'https://www.narev.ai/api/models/pricing' \
  --data-urlencode 'model_id=llama-3.1-70b'
```

```bash
# Paginate the full catalog
curl -G 'https://www.narev.ai/api/models/pricing' \
  --data-urlencode 'limit=1000' \
  --data-urlencode 'page=2'
```

--------------------------------

### Create Offline Price Resolver with Snapshot

Source: https://context7.com/narevai/skills/llms.txt

Integrates a pricing snapshot into `@ai-billing/core` for offline billing, avoiding runtime HTTP calls to Narev. Ensure the pricing map uses USD per token, not per million.

```typescript
// src/billing/price-resolver.ts
// After running the snapshot script, use createObjectPriceResolver
// for offline billing — no runtime HTTP call to Narev.
// NOTE: SDK ModelPricing expects USD per token (NOT per million). Do not
// apply the 1_000_000 multiplier when building this map.

import { createObjectPriceResolver, type ModelPricing } from "@ai-billing/core";

// Map from Narev API (snake_case, USD/token) → SDK (camelCase, USD/token)
async function buildPricingMap(): Promise<Record<string, ModelPricing>> {
  const res = await fetch("https://www.narev.ai/api/models/pricing?limit=1000");
  if (!res.ok) process.exit(1);
  const { data } = await res.json();

  const map: Record<string, ModelPricing> = {};
  for (const row of data) {
    if (!row.pricing) continue;
    const p = row.pricing;
    map[`${row.provider}/${row.model_id}`] = {
      promptTokens:            p.price_prompt             ?? 0,
      completionTokens:        p.price_completion         ?? 0,
      inputCacheReadTokens:    p.price_input_cache_read   ?? 0,
      inputCacheWriteTokens:   p.price_input_cache_write  ?? 0,
      internalReasoningTokens: p.price_internal_reasoning ?? 0,
      request:                 p.pricing_request          ?? 0,
      webSearch:               p.price_web_search         ?? 0,
      discount:                p.pricing_discount         ?? 0,
    };
  }
  return map;
}

export const priceResolver = createObjectPriceResolver(await buildPricingMap());

// Usage in app:
// import { priceResolver } from "./billing/price-resolver";
// const billing = createBilling({ priceResolver, destination: myDestination });

```

--------------------------------

### Create Object Price Resolver with Local Pricing Data

Source: https://github.com/narevai/skills/blob/main/skills/pricing/narev-update-llm-pricing/SKILL.md

Use this pattern when your application already uses the Narev SDK and you have pricing data in a local object. Ensure the pricing data is in USD per token, not per million tokens.

```typescript
import { createObjectPriceResolver } from "@ai-billing/core";
import { pricing } from "./pricing";

export const priceResolver = createObjectPriceResolver(pricing);
```

--------------------------------

### Narev Skills Directory Structure

Source: https://github.com/narevai/skills/blob/main/CLAUDE.md

Illustrates the standard directory layout for Narev AI agent skills, separating core routing logic from pricing-specific functionalities.

```tree
skills/
├── core/
│   └── narev/                     # router — when to use which skill or doc path
└── pricing/
    ├── narev-lookup-llm-pricing/  # Pricing API reference (GET catalog, POST calculate)
    └── narev-update-llm-pricing/  # Snapshot / pin rates into the repo
```

--------------------------------

### Calculate Model Pricing

Source: https://context7.com/narevai/skills/llms.txt

This endpoint calculates the cost for a given model usage based on prompt tokens, completion tokens, and other usage parameters. It supports various providers like OpenAI and Anthropic.

```APIDOC
## POST /api/models/pricing/calculate

### Description
Calculates the cost for a specified model usage, including prompt tokens, completion tokens, and cache usage.

### Method
POST

### Endpoint
https://www.narev.ai/api/models/pricing/calculate

### Parameters
#### Request Body
- **modelId** (string) - Required - The identifier of the model.
- **provider** (string) - Required - The provider of the model (e.g., "openai", "anthropic").
- **subprovider** (string) - Optional - The subprovider for multi-host models (e.g., "bedrock", "openrouter").
- **usage** (object) - Required - An object containing token usage details.
  - **promptTokens** (integer) - Required - The number of prompt tokens used.
  - **completionTokens** (integer) - Required - The number of completion tokens used.
  - **cacheReadTokens** (integer) - Required - The number of cache read tokens used.
  - **cacheWriteTokens** (integer) - Required - The number of cache write tokens used.
  - **reasoningTokens** (integer) - Required - The number of reasoning tokens used.

### Request Example
```json
{
  "modelId": "gpt-4o",
  "provider": "openai",
  "subprovider": "OpenAI",
  "usage": {
    "promptTokens": 1000,
    "completionTokens": 500,
    "cacheReadTokens": 0,
    "cacheWriteTokens": 0,
    "reasoningTokens": 0
  }
}
```

### Response
#### Success Response (200)
- **pricing** (object) - Contains pricing details per usage type.
  - **input** (number) - Cost per input token.
  - **output** (number) - Cost per output token.
  - **request** (number | null) - Cost per request.
  - **inputCacheRead** (number) - Cost per input cache read token.
  - **inputCacheWrite** (number | null) - Cost per input cache write token.
  - **internalReasoning** (number | null) - Cost per internal reasoning token.
  - **webSearch** (number | null) - Cost per web search token.
- **costBreakdown** (object)
  - **total** (number) - The total calculated cost.
- **usage** (object) - Echoes the usage object from the request.

#### Response Example
```json
{
  "pricing": {
    "input": 0.0000025,
    "output": 0.00001,
    "request": null,
    "inputCacheRead": 0.00000125,
    "inputCacheWrite": null,
    "internalReasoning": null,
    "webSearch": null
  },
  "costBreakdown": {
    "total": 0.0000075
  },
  "usage": {
    "promptTokens": 1000,
    "completionTokens": 500,
    "cacheReadTokens": 0,
    "cacheWriteTokens": 0,
    "reasoningTokens": 0
  }
}
```
```

--------------------------------

### narev-lookup-llm-pricing: List Model Pricing

Source: https://context7.com/narevai/skills/llms.txt

Retrieves the pricing catalog for LLM models with optional filters for model ID and provider. Supports pagination and returns pricing details per token.

```APIDOC
## GET /api/models/pricing — List the model catalog

### Description
Retrieves the pricing catalog with optional filters. Supports pagination (`page`, `limit` up to 1000). Each entry contains `model_id`, `provider`, `subprovider`, and a `pricing` object with per-token rates.

### Method
GET

### Endpoint
/api/models/pricing

### Parameters
#### Query Parameters
- **model_id** (string) - Optional - The ID of the model to filter by.
- **provider** (string) - Optional - The provider of the model to filter by.
- **subprovider** (string) - Optional - The subprovider of the model to filter by, required when the model is hosted by multiple vendors.
- **limit** (integer) - Optional - The maximum number of results to return (default 100, max 1000).
- **page** (integer) - Optional - The page number for pagination.

### Request Example
```bash
# List pricing for gpt-4o on OpenAI
curl -G 'https://www.narev.ai/api/models/pricing' \
  --data-urlencode 'model_id=gpt-4o' \
  --data-urlencode 'provider=openai'

# List all providers for a multi-host model (subprovider matters)
curl -G 'https://www.narev.ai/api/models/pricing' \
  --data-urlencode 'model_id=llama-3.1-70b'

# Paginate the full catalog
curl -G 'https://www.narev.ai/api/models/pricing' \
  --data-urlencode 'limit=1000' \
  --data-urlencode 'page=2'
```

### Response
#### Success Response (200)
- **data** (array) - An array of pricing objects.
  - **model_id** (string) - The ID of the model.
  - **provider** (string) - The provider of the model.
  - **subprovider** (string) - The subprovider of the model.
  - **pricing** (object) - An object containing per-token rates.
    - **price_prompt** (number) - The price for prompt tokens.
    - **price_completion** (number) - The price for completion tokens.
    - **price_input_cache_read** (number) - The price for input cache read.
    - **price_input_cache_write** (number | null) - The price for input cache write.
    - **price_internal_reasoning** (number | null) - The price for internal reasoning.
    - **pricing_request** (number | null) - The price for pricing requests.
    - **price_web_search** (number | null) - The price for web search.
    - **pricing_discount** (number) - The pricing discount.
- **meta** (object) - Metadata for pagination.
  - **page** (integer) - The current page number.
  - **limit** (integer) - The number of results per page.
  - **total** (integer) - The total number of results.
  - **total_pages** (integer) - The total number of pages.

### Response Example
```json
{
  "data": [
    {
      "model_id": "gpt-4o",
      "provider": "openai",
      "subprovider": "OpenAI",
      "pricing": {
        "price_prompt": 0.0000025,
        "price_completion": 0.00001,
        "price_input_cache_read": 0.00000125,
        "price_input_cache_write": null,
        "price_internal_reasoning": null,
        "pricing_request": null,
        "price_web_search": null,
        "pricing_discount": 0
      }
    }
  ],
  "meta": { "page": 1, "limit": 100, "total": 1, "total_pages": 1 }
}
```
```

--------------------------------

### List LLM Pricing Catalog

Source: https://github.com/narevai/skills/blob/main/skills/pricing/narev-lookup-llm-pricing/SKILL.md

Retrieves a paginated list of available LLM models and their pricing details. Supports filtering by model ID, provider, and search terms.

```APIDOC
## GET /api/models/pricing

### Description
Lists the catalog of LLM models and their associated pricing information. This endpoint is paginated and can be filtered by various parameters.

### Method
GET

### Endpoint
https://www.narev.ai/api/models/pricing

### Parameters
#### Query Parameters
- **model_id** (string) - Optional - Filter by specific model ID.
- **search** (string) - Optional - Perform a general search across models.
- **provider** (string) - Optional - Filter by the model provider (e.g., 'openai').
- **subprovider** (string) - Optional - Filter by a specific subprovider.
- **sort_by** (string) - Optional - Field to sort the results by (`model_id`, `provider`, or `subprovider`).
- **order** (string) - Optional - Order of sorting (`asc` or `desc`).
- **page** (integer) - Optional - Page number for pagination (default: 1).
- **limit** (integer) - Optional - Number of results per page (max: 1000, default: 100).

### Response
#### Success Response (200)
- **data** (array) - An array of `ModelPricingEntry` objects.
  - **model_id** (string) - The ID of the model.
  - **provider** (string) - The provider of the model.
  - **subprovider** (string) - The subprovider of the model.
  - **pricing** (object) - Pricing details for the model.
    - **price_prompt** (number) - USD per input token.
    - **price_completion** (number) - USD per output token.
    - **price_input_cache_read** (number) - USD per cached input token.
    - **price_input_cache_write** (number) - USD per cached input token.
    - **price_internal_reasoning** (number) - USD per reasoning output token.
    - **pricing_request** (number) - Flat USD per request.
    - **price_web_search** (number) - USD per web-search invocation.
    - **pricing_discount** (number) - Fractional discount (0-1).
    - **price_image** (number) - USD per image unit (if applicable).
    - **price_image_output** (number) - USD per output image unit (if applicable).
    - **price_audio** (number) - USD per audio unit (if applicable).
    - **price_audio_output** (number) - USD per output audio unit (if applicable).
    - **price_input_audio_cache** (number) - USD per input audio cache unit (if applicable).
- **meta** (object) - Pagination metadata.
  - **page** (integer) - Current page number.
  - **limit** (integer) - Results per page.
  - **total** (integer) - Total number of results.
  - **total_pages** (integer) - Total number of pages.

### Request Example
```bash
curl -G 'https://www.narev.ai/api/models/pricing' \
  --data-urlencode 'model_id=gpt-4o' \
  --data-urlencode 'provider=openai'
```
```

--------------------------------

### Calculate LLM Call Cost with cURL

Source: https://github.com/narevai/skills/blob/main/skills/pricing/narev-lookup-llm-pricing/SKILL.md

Use this cURL command to send a POST request to the Narev pricing calculation endpoint. Ensure all required usage fields are provided as integers, using 0 for unused categories.

```bash
curl -X POST 'https://www.narev.ai/api/models/pricing/calculate' \
  -H 'Content-Type: application/json' \
  -d '{
    "modelId": "gpt-4o",
    "provider": "openai",
    "subprovider": "OpenAI",
    "usage": {
      "promptTokens": 1000,
      "completionTokens": 500,
      "cacheReadTokens": 0,
      "cacheWriteTokens": 0,
      "reasoningTokens": 0
    }
  }'
```

--------------------------------

### Codex Plugin Manifest

Source: https://github.com/narevai/skills/blob/main/AGENTS.md

Specifies the Codex plugin manifest for bundling Narev skills, with the 'skills' entry point directing to the './skills/' directory.

```json
.codex-plugin/plugin.json — Codex plugin manifest for the bundle (`skills` entry point at `./skills/`).
```

--------------------------------

### narev-lookup-llm-pricing: Calculate Call Cost

Source: https://context7.com/narevai/skills/llms.txt

Computes the exact USD cost for a single LLM call based on provided token usage. Requires all five usage fields and the subprovider if applicable.

```APIDOC
## POST /api/models/pricing/calculate — Calculate USD cost for one call

### Description
Computes the exact USD cost for a single LLM call given token usage. All five `usage` integer fields are required — pass `0` for unused categories (never `null` or omit them). `subprovider` is required when the model is served by multiple hosts.

### Method
POST

### Endpoint
/api/models/pricing/calculate

### Parameters
#### Request Body
- **usage** (object) - Required - An object containing token usage for different categories.
  - **prompt_tokens** (integer) - Required - The number of prompt tokens used.
  - **completion_tokens** (integer) - Required - The number of completion tokens used.
  - **input_cache_read_tokens** (integer) - Required - The number of input cache read tokens used.
  - **input_cache_write_tokens** (integer) - Required - The number of input cache write tokens used.
  - **internal_reasoning_tokens** (integer) - Required - The number of internal reasoning tokens used.
- **model_id** (string) - Required - The ID of the model used.
- **provider** (string) - Required - The provider of the model.
- **subprovider** (string) - Optional - The subprovider of the model, required when the model is hosted by multiple vendors.

### Request Example
```json
{
  "usage": {
    "prompt_tokens": 1000,
    "completion_tokens": 500,
    "input_cache_read_tokens": 200,
    "input_cache_write_tokens": 0,
    "internal_reasoning_tokens": 0
  },
  "model_id": "gpt-4o",
  "provider": "openai",
  "subprovider": "OpenAI"
}
```

### Response
#### Success Response (200)
- **total_cost_usd** (number) - The total cost of the call in USD.

#### Response Example
```json
{
  "total_cost_usd": 0.00000425
}
```
```

--------------------------------

### Calculate LLM Call Cost

Source: https://github.com/narevai/skills/blob/main/skills/pricing/narev-lookup-llm-pricing/SKILL.md

Computes the estimated USD cost for a single LLM API call based on the model, provider, and usage details.

```APIDOC
## POST /api/models/pricing/calculate

### Description
Calculates the estimated USD cost for a single LLM API call. You need to provide the model ID, provider, and a breakdown of token usage.

### Method
POST

### Endpoint
https://www.narev.ai/api/models/pricing/calculate

### Parameters
#### Request Body
- **modelId** (string) - Required - The ID of the model to use.
- **provider** (string) - Required - The provider of the model (e.g., 'openai').
- **subprovider** (string) - Required - The subprovider, especially when a model is served by multiple providers (e.g., 'bedrock', 'openrouter').
- **usage** (object) - Required - An object detailing token usage.
  - **promptTokens** (integer) - Required - Number of input tokens.
  - **completionTokens** (integer) - Required - Number of output tokens.
  - **cacheReadTokens** (integer) - Required - Number of cached input tokens read.
  - **cacheWriteTokens** (integer) - Required - Number of cached input tokens written.
  - **reasoningTokens** (integer) - Required - Number of reasoning output tokens.
- **webSearchCount** (integer) - Optional - Number of web searches performed.
- **isByok** (boolean) - Optional - Indicates if Bring Your Own Key is used.

### Request Example
```json
{
  "modelId": "claude-3-sonnet-20240229",
  "provider": "anthropic",
  "subprovider": "bedrock",
  "usage": {
    "promptTokens": 1000,
    "completionTokens": 500,
    "cacheReadTokens": 0,
    "cacheWriteTokens": 0,
    "reasoningTokens": 100
  },
  "webSearchCount": 1,
  "isByok": false
}
```

### Response
#### Success Response (200)
- **estimatedCostUsd** (number) - The calculated cost in USD for the API call.

#### Response Example
```json
{
  "estimatedCostUsd": 0.0012345
}
```
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.