> ## Documentation Index > Fetch the complete documentation index at: https://docs.hyperline.co/llms.txt > Use this file to discover all available pages before exploring further. # Create a custom agent cost item > Price a non-LLM cost item the registry has never heard of — a search API, a vector DB, GPU seconds, a render. The item is priced entirely by its `dimensions` (its token rates are 0). `provider` and `item` must match, verbatim, the `vendor` and `item` your SDK sends on `hyperline.cost({ vendor, item, quantity, unit })` — they are lowercased and trimmed, nothing else — and each `dimensions[].key` must match the `unit` of the call it prices: `cost()` writes the quantity to `usage_extra[unit]` and pricing matches a dimension by that exact key, so a key that does not match prices nothing. Idempotent by design so a seed script can be re-run: an open custom item already standing for that (provider, item) is returned as it is, rates untouched, with a 200 instead of a fresh 201 — edit a rate by deleting the item and creating it again. A (provider, item) pair the global registry already prices is refused: that model has a price, and a negotiated rate for it goes through the overrides endpoint. ## OpenAPI ````yaml post /v1/agents/cost-catalog/items openapi: 3.1.0 info: title: Hyperline API version: 0.0.0 servers: - url: https://api.hyperline.co - url: https://sandbox.api.hyperline.co security: [] paths: /v1/agents/cost-catalog/items: post: tags: - Agents summary: Create a custom agent cost item description: >- Price a non-LLM cost item the registry has never heard of — a search API, a vector DB, GPU seconds, a render. The item is priced entirely by its `dimensions` (its token rates are 0). `provider` and `item` must match, verbatim, the `vendor` and `item` your SDK sends on `hyperline.cost({ vendor, item, quantity, unit })` — they are lowercased and trimmed, nothing else — and each `dimensions[].key` must match the `unit` of the call it prices: `cost()` writes the quantity to `usage_extra[unit]` and pricing matches a dimension by that exact key, so a key that does not match prices nothing. Idempotent by design so a seed script can be re-run: an open custom item already standing for that (provider, item) is returned as it is, rates untouched, with a 200 instead of a fresh 201 — edit a rate by deleting the item and creating it again. A (provider, item) pair the global registry already prices is refused: that model has a price, and a negotiated rate for it goes through the overrides endpoint. operationId: createAgentCostCatalogItem requestBody: description: Create custom cost item payload content: application/json: schema: type: object properties: provider: type: string minLength: 1 item: type: string minLength: 1 dimensions: type: array items: type: object properties: key: type: string pattern: ^[a-z0-9_.]{1,64}$ rate: type: string pattern: ^\d+(\.\d+)?$ per: type: string enum: - 1m - 1k - unit required: - key - rate - per minItems: 1 maxItems: 10 required: - provider - item - dimensions responses: '200': description: The custom cost item already existed content: application/json: schema: type: object properties: provider: type: string model: type: string aliases: type: array items: type: string input_per_m_tokens: type: string output_per_m_tokens: type: string cached_input_per_m_tokens: type: string cache_write_per_m_tokens: type: string dimensions: type: array items: type: object properties: key: type: string rate: type: string per: type: string enum: - 1m - 1k - unit required: - key - rate - per effective_from: type: string source: type: string custom_item_id: type: - string - 'null' override: type: - object - 'null' properties: id: type: string input_per_m_tokens: type: string output_per_m_tokens: type: string cached_input_per_m_tokens: type: - string - 'null' cache_write_per_m_tokens: type: - string - 'null' dimensions: type: array items: type: object properties: key: type: string rate: type: string per: type: string enum: - 1m - 1k - unit required: - key - rate - per effective_from: type: string required: - id - input_per_m_tokens - output_per_m_tokens - cached_input_per_m_tokens - cache_write_per_m_tokens - effective_from required: - provider - model - aliases - input_per_m_tokens - output_per_m_tokens - cached_input_per_m_tokens - cache_write_per_m_tokens - dimensions - effective_from - source - custom_item_id - override '201': description: '' content: application/json: schema: type: object properties: provider: type: string model: type: string aliases: type: array items: type: string input_per_m_tokens: type: string output_per_m_tokens: type: string cached_input_per_m_tokens: type: string cache_write_per_m_tokens: type: string dimensions: type: array items: type: object properties: key: type: string rate: type: string per: type: string enum: - 1m - 1k - unit required: - key - rate - per effective_from: type: string source: type: string custom_item_id: type: - string - 'null' override: type: - object - 'null' properties: id: type: string input_per_m_tokens: type: string output_per_m_tokens: type: string cached_input_per_m_tokens: type: - string - 'null' cache_write_per_m_tokens: type: - string - 'null' dimensions: type: array items: type: object properties: key: type: string rate: type: string per: type: string enum: - 1m - 1k - unit required: - key - rate - per effective_from: type: string required: - id - input_per_m_tokens - output_per_m_tokens - cached_input_per_m_tokens - cache_write_per_m_tokens - effective_from required: - provider - model - aliases - input_per_m_tokens - output_per_m_tokens - cached_input_per_m_tokens - cache_write_per_m_tokens - dimensions - effective_from - source - custom_item_id - override security: - bearer: [] components: securitySchemes: bearer: type: http scheme: bearer bearerFormat: JWT ````