Adapter manifest

Status: Genesis implementation format; not yet a stable protocol RFC. A manifest declares:

adapter identity and version;
origin scope;
read operation and resource type;
semantic module closure;
source-native fields;
JSON Pointer locators;
output types;
required or optional status;
deterministic transformations;
mapping relation.

{
  "id": "price_currency",
  "native_term": "origin:testorigin/currency",
  "semantic_term": "commerce:OfferPrice.currency",
  "json_pointer": "/offer/currency",
  "value_type": "currency_code",
  "required": true,
  "transforms": ["trim", "uppercase"],
  "mapping_relation": "equivalent_in_context"
}

Failure behavior

Missing required field: extraction fails closed.
Missing optional field: output status is unresolved and still carries provenance.
An empty source string is a resolved value and remains distinct from unresolved.
Invalid decimal or currency code: extraction fails.
Origin mismatch: adapter refuses execution.
Non-JSON media type: Genesis JSON adapter refuses execution.
Duplicate object keys, trailing JSON values, and unknown manifest properties are rejected.

Bounded JSON profile

Both manifests and observed representations must be valid UTF-8 and contain exactly one top-level JSON value followed only by whitespace. Duplicate object keys are rejected at every depth. Escaped UTF-16 surrogate code units must form valid pairs; lone surrogates are rejected rather than replaced with a different lexical value.

Limit	Manifest	Observed representation
Document bytes	262,144	Observation body limit, at most 2,097,152
Nested containers	32	64
Decoded scalar bytes	8,192	65,536
Entries in one object or array	4,096	16,384
JSON tokens	32,768	65,536

A manifest contains at most 64 fields and 64 semantic modules. A field contains at most 16 transforms. A JSON Pointer is at most 4,096 bytes and 128 tokens. Pointer escapes follow RFC 6901; array indexes are 0 or an unsigned decimal without leading zeroes.

Lexical preservation and transforms

For a JSON scalar, native.lexical_value is its decoded lexical value. JSON number spelling is preserved. The exact original JSON token remains available in the immutable body through the body digest and JSON Pointer locator. trim, uppercase, and lowercase operate on ASCII only. They do not apply implicit Unicode normalization or locale-sensitive case conversion. The native value is recorded before every declared transform; the transformed value appears only in the semantic view. Serialized results are bounded to 8 MiB and are published by a synchronized temporary file followed by atomic rename. A partial result is never promoted to its final path.

Digest behavior

The current adapter digest hashes the exact manifest bytes. A later canonical adapter-package format will define deterministic structured encoding and signatures.

Start

Core concepts

Architecture

Protocol

Security

Governance

Contribute

Failure behavior

Bounded JSON profile

Lexical preservation and transforms

Digest behavior

​Failure behavior

​Bounded JSON profile

​Lexical preservation and transforms

​Digest behavior

Failure behavior

Bounded JSON profile

Lexical preservation and transforms

Digest behavior