Home Docs Schema Formats Errors
Conduit

Core

Philosophy Getting Started Roadmap

Formats

Formats ▶
Overview JSON XML YAML TOML NDJSON CSV CBOR

Configuration

Configuration ▶
Overview Server Settings Database Settings Policy Settings Authorization Overview Environment Auth Database Auth No-Op Auth

API Reference

API Reference ▶
Schema API CRUD API Validation Pagination

Architecture

Architecture ▶
Overview High-Level Flow Multi-Format Engine Database Drivers Error System

Errors

Errors ▶
Error Index 400 Bad Request 401 Unauthorized 403 Forbidden 404 Not Found 405 Method Not Allowed 409 Conflict 415 Unsupported Media Type 422 Unprocessable Entity 429 Too Many Requests 500 Internal Server Error 501 Not Implemented 503 Service Unavailable

Multi-Format Engine

Multi-Format Engine

Conduit includes a full multi-format engine capable of decoding incoming payloads and encoding responses into a wide range of formats. This allows clients to interact with the API using whichever format best fits their environment or tooling.

Multi-Format Engine

Conduit decodes input formats (JSON, XML, YAML, TOML, NDJSON, CSV, CBOR) into internal structures and encodes responses back into the requested format. Format negotiation is automatic and based on the Content-Type and Accept headers or the ?format= query parameter.

Input Formats

Conduit supports the following input formats for POST, PUT, and PATCH:

  • JSON
  • YAML
  • TOML
  • XML
  • NDJSON (streamed objects)
  • CSV (header-based mapping)
  • CBOR (binary structured data)

The decoder automatically selects the correct parser based on the Content-Type header. Unsupported formats return a structured error.

Output Formats

All endpoints support multiple output formats:

  • JSON
  • YAML
  • XML
  • TOML
  • CSV
  • NDJSON
  • CBOR

Output format is selected using either the Accept header or the ?format= query parameter. 

GET /v1/books?format=xml
GET /v1/books?format=toml
GET /v1/books?format=ndjson

Format Negotiation

Conduit determines the output format using the following priority:

  1. ?format= query parameter
  2. Accept header
  3. Default: JSON

If the requested format is unsupported, Conduit returns a structured error describing the available formats.

NDJSON Streaming

NDJSON (Newline‑Delimited JSON) is ideal for large datasets and streaming pipelines. Conduit outputs one JSON object per line and flushes the response incrementally when supported by the HTTP server. 

{"id":1,"title":"Example"}
{"id":2,"title":"Another"}

NDJSON input is also supported for bulk inserts.

CSV

CSV output includes a header row followed by one row per record. Column order is determined by the schema. 

id,title,author
1,Foundation,Isaac Asimov
2,Dune,Frank Herbert

CSV input maps header names to column names. Missing columns are ignored; extra columns are ignored.

CBOR

CBOR is a compact binary format ideal for IoT devices and high‑performance clients. Conduit uses a deterministic CBOR encoder to ensure consistent output across requests.

CBOR input is parsed into the same internal structure as JSON, YAML, XML, and TOML.

Internal Model

Regardless of input format, Conduit normalizes all payloads into a unified internal structure. This ensures consistent validation, authorization, and database behavior across formats.

The encoder then converts the internal structure into the requested output format.

Notes

  • All formats support schema and CRUD endpoints.
  • NDJSON is ideal for streaming large datasets.
  • CSV input requires a header row.
  • CBOR is recommended for embedded systems.
  • Format negotiation is consistent across all endpoints.
  • Errors are encoded in the requested format whenever possible.