 | This open source module is sponsored and supported by Voxgig. |
|---|
Potencialize seu agente
O comando do próprio publicador. Não é um endpoint do PlugMCP.
Host a copy with our keys and logs.
npx -y [email protected]O Claude Code não tem link de instalação para um comando local. Copie o comando abaixo e execute-o em um terminal.
claude mcp add aontu -- npx -y aontu@0.76.3Execute em um terminal.
{ "mcpServers": { "aontu": { "command": "npx", "args": [ "-y", "[email protected]" ] } }}claude_desktop_config.json
URL remota ou um comando na sua máquina? Como as duas instalações diferem
Rodar no PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
Abrir | This open source module is sponsored and supported by Voxgig. |
|---|
aontu combines data, schemas, and defaults into one consistent result, or reports where they conflict. It is a purpose-specific dialect inspired by CUE.
aontu ships two implementations, kept in parity (structure inspired by
voxgig/util):
Both are checked against a single, language-agnostic test suite in
test/spec/ (tab-separated cases run by both
implementations). See AGENTS.md and
docs/shared-spec.md.
make build # build both (ts + go)
make test # test both against the shared spec
make install # put both builds on PATH from this clone (npm link, go install)
ts/ canonical TypeScript implementation (src, test, dist, dist-test)
go/ Go port (package aontu)
test/spec/ shared *.tsv unit tests both implementations must satisfy
docs/ documentation
Both implementations ship an aontu command that evaluates a file (or
stdin) and prints the result, or starts a REPL when run with no file:
aontu config.aontu # evaluate a file -> JSON
aontu --canon config.aontu # canonical form instead
echo 'a:1 b:$.a' | aontu # read from stdin
aontu # no file on a terminal -> REPL
Beyond evaluation the command has a verb for each task around a
document: vet (validate data against a schema), model get and
model why (query, and provenance), model set (change a value in an
overlay, by appending or in place), subsume and breaking (schema evolution),
hash (pin what a document means), relations, reaches and view
(the declared entity graph: its checks, reachability over it, and its
figures drawn as text: tree, matrix, graph, layers, sets, the meet
ladder, the subsumption poset, the key document and the value
lattice), jsonschema (export
the model as JSON Schema), trace (what wrote a line of generated
output: the file, the model node and the rule), template (read a
generator written in the target's own syntax), trim (find redundant
entries), sync, add, get, remove, why and publish (the
package system: dependency closures and their repository), agentsmd (an AGENTS.md stanza), fmt (the
source formatter, in the tradition of gofmt: one agreed form), lsp
(the language server) and mcp (the MCP server, npm build). The
library adds a path-addressed diff, and an MCP server answers with
the same reports over stdio. All of it is documented in
docs/reference-api.md.
Install with npm i -g aontu (Node); with
curl -fsSL https://aontu.dev/install.sh | sh on Linux or macOS, which
puts the Go build of aontu and aontu-lsp in ~/.local/bin with no
toolchain; with a package or archive from the
releases page, the
image ghcr.io/aontu-lang/aontu, the Nix flake, or the setup action
aontu-lang/aontu/setup-action in a workflow; or with
go install github.com/aontu-lang/aontu/go/cmd/aontu@latest (Go). From a
clone, make install puts both builds on PATH: make install-ts links
the checkout as the global npm package and make install-go runs
go install for the two commands. Without installing:
node ts/bin/aontu.js … or, inside go/, go run ./cmd/aontu ….
aontu fmt writes a document in one agreed form, in the tradition of
gofmt: layout stops being an argument, and a diff shows only what
changed. The form is a spelling and never a change of meaning. What
comes back evaluates to the same value, carries the same canon-hash,
and formatting it again changes nothing.
aontu fmt config.aontu # print the formatted text
aontu fmt -w config.aontu # rewrite in place
aontu fmt --check *.aontu # exit 1 if any file would change: the CI gate
aontu fmt -d config.aontu # a unified diff of what would change
aontu fmt --lint config.aontu # style findings only; the form is untouched
The rules, in short:
LF endings, no trailing
whitespace, one final newline.key: value, the colon tight to the key. No commas between
entries; a call keeps its argument commas.a: b: c: 1; a one-key map in a list is a pair element,
[a:1 b:2]. A map that is an operand or an argument keeps its
braces, because splitting it would change the document.server: host: …
over server: port: …. A key written twice is a meet, so the two
spellings are one document.A file as it arrived from JSON:
{
"server": { "host": '0.0.0.0', "port": 8080 },# where the edge listens
"limits": { "rps": 100, "burst": 200 },
"features": ["auth", "metrics"]
}
aontu fmt writes:
server: { host:"0.0.0.0" port:8080 } # where the edge listens
limits: { rps:100 burst:200 }
features: ["auth" "metrics"]
Both ports expose the formatter as a library, with the same report:
import { format, unifiedDiff } from 'aontu'
const r = format('a: 1# c\n', { lint: true })
r.verdict // 'formatted'
r.changed // true
r.text // 'a: 1 # c\n'
r.findings // the --lint style findings, [] here
unifiedDiff('a.aontu', 'a: 1# c\n', r.text)
r := aontu.New().Format("a: 1# c\n")
r.Verdict // "formatted"
r.Changed // true
r.Text // "a: 1 # c\n"
aontu.New().FormatWith(src, aontu.FormatOptions{Lint: true})
aontu.UnifiedDiff("a.aontu", src, r.Text)
The full rule set is the formatted form; the verb, its options and its exit codes are in the API reference; running it on a repository and gating CI on it is a how-to.
Full documentation is in docs/:
use-cases/ holds eighteen enterprise-shaped systems built
as real aontu documents, each with a check.sh that drives the CLI and
asserts every outcome.
JSON-superset unification language and validation gate for AI-agent ground truth - evaluate, vet, diff, query, and evolution-check structured models.
{ "mcpServers": { "aontu": { "command": "npx", "args": [ "-y", "[email protected]" ] } }}Ferramentas
Ainda não há lista de ferramentas. Servidores remotos recebem uma após a primeira verificação de saúde.
Perguntas frequentes
Copie nesta página a configuração stdio para Cursor, Claude Desktop, Claude Code ou VS Code. Ela usa a instalação do publicador, não um proxy do PlugMCP.
Este registro não lista nenhuma variável de ambiente obrigatória.
Aontu roda localmente via stdio.
Última verificação: 2026-11-15. JSON-superset unification language and validation gate for AI-agent ground truth - evaluate, vet, diff, query, and evolution-check structured models.