Docs
/
adapter system

a-transformation

Input-typed transformation registry: 55 adapters keyed "<inputType>.<name>" across 26 input types (string, hex, base64, bytes, json, yaml, csv, url, html, markdown, code, env, geo, vector...

a-transformation (adapter system)

Input-typed transformation registry. Ported from leumas-middleware/lib/a.transformation — 52 adapters across 26 input types. Stripped: server.js + static server, frontend/, src/cli, src/server, src/core runner/registry (the SDK/CLI/HTTP shells), bin/, fixtures/.

Each source adapter (export async function run(input, ctx) + meta.json) is preserved under ./adapters/<inputType>/<name>/. Tools are keyed "<inputType>.<name>" and called as adapters[id] = (args) => run(args.input, { options: args.options ?? args }). All 52 use only Node builtins — no external dependencies.

Port fixes: deprecated import assertions (assert { type: 'json' }) rewritten to import attributes (with { type: 'json' }); code.stripComments had corrupted regex literals (//[sS]?*//g) — fixed to correct block/line-comment stripping.

Input types (26)

archive, audio, base64, bytes, cli-args, code, config, csv, dataset, env, files, folders, geo, hex, html, image, json, markdown, stream, string, timestamps, url, vector, video, webhook, yaml.

Examples

toolargsresult
string.slugify{ input: 'Hello World!', options: { separator: '-' } }"hello-world"
string.base64{ input: 'hi' }base64 string
hex.normalize{ input: '0xAA bb cc' }normalized hex
json.parse / json.stringify{ input }parsed / stringified
code.stripComments{ input: 'a//x\nb/y/c' }comment-free source

The exported inputTypes array lists all input-type groups. Each adapter's meta.json (options / inputFields / returns) lives beside its adapter.js.

Source shared/engines/adapters/domain/a-transformation/README.md (no-git)markdownjson
Generated from the Leumas repository. Every page cites the file it came from.leumas.techllms.txt