Cette page a été traduite automatiquement. Lire l’original en anglais.

Component

core/ai-agent/librechat

A self-hosted LibreChat installation — the browser platform a person holds conversations in, and the tier that turns a bare model endpoint into a working assistant surface. It reaches its model through the `model` uses-socket and adds agents an author assembles without writing code, tool invocation over the tool servers an administrator registers, a code interpreter, and retrieval over files a user uploads. Every agent run, a delegated child run included, is entered from a conversation a person started: the platform runs no background loop of its own. Its accounts, conversations, agents and stored per-user credentials live in a document store wired through the `backend` socket, the system of record; only uploaded files sit on a volume of its own. A store that requires a sign-in, as FerretDB does by default, takes the user and password from the credential attached on that connection. The two token-signing secrets and the credential-encryption pair decide what its sign-ins and stored credentials are worth, and each has a slot of its own. None of the four has a default, and the build halts until each slot is filled. On Kubernetes, reach for `librechat-chart`, rendering the project's own chart, where a chart repository is reachable at build time, and for `librechat`, the workload synthesis, where the build must fetch nothing. With `values.metrics_enabled` true each Recipe turns on the platform's Prometheus metrics, at /metrics on its port and answered only to a bearer token equal to the secret attached to metrics_secret, and publishes the address as `metrics_endpoint`; a scrape presents that secret on its connection's `credential` slot.

Version

Provenance

OfficielSigné par une identité de signature que ce catalogue reconnaît comme éditeur officiel.
Empreinte du contenu
sha256:df8acd69b138eb9e74cf3bf6b7819fe1bc5350bc21a667c476305317f2d5fe24
Publié
2026-11-15T08:08:55Z
Identité de signature
https://token.actions.githubusercontent.com/https://github.com/aggrega-ai/aggrega-hub/.github/workflows/catalog-publish-sign.yml@refs/heads/main

Signature

  • Officielapplication/vnd.dev.sigstore.bundle.v0.3+json

    https://token.actions.githubusercontent.com/https://github.com/aggrega-ai/aggrega-hub/.github/workflows/catalog-publish-sign.yml@refs/heads/main

    sha256:ce3140ba588b7ddc4c67693ef09346ac88b32a6c5db49609981eb6a0d03e753e

Référence à cette version

oci://hub.aggrega.ai/component/core/ai-agent/librechat:1.2.0

Paramètres du développeur

Valeurs que le développeur définit dans l’Aggregate.

Étend BaseComponents#/schemas/BaseAgentParameter

NomTypeValeur par défautObligatoireDescription
app_titlestringLibreChatnonName the platform presents in the browser title and the interface header. Two installations reachable by the same people tell themselves apart by carrying different titles.
database_namestringLibreChatnonName of the logical database the platform uses inside the wired document store. Two installations sharing one store keep their data apart by carrying different names.
allow_registrationbooleanfalsenonWhether a visitor who is not yet a user may create an account from the sign-in page. Left off, accounts are created only by an administrator, which is the posture an installation reachable beyond its own team wants.
metrics_enabledbooleanfalsenonTurns on LibreChat's own metrics, in the Prometheus format, at /metrics on the platform's port, and publishes their address as `metrics_endpoint`; while false that key is empty. The platform answers them only to the secret attached to metrics_secret, presented as a bearer token, so the build halts when this is true and that slot is unfilled. They carry request counts and timings, never a user's or conversation's identifier or content.

Dimensionnement de l’opérateur

Dimensionnement de l’infrastructure que l’opérateur ajuste pour chaque environnement.

NomTypeValeur par défautObligatoireDescription
portinteger (min 1024, max 65535)3080nonListen port the platform serves its interface and API on, and the port consumers connect to. Most operators should omit this and let the Recipe assign the default (3080).
published_portinteger (min 1024, max 65535)—nonPort published on the host's loopback address, 127.0.0.1, on Docker Compose, forwarding to infra.port inside the container; any process on this machine can reach it there. Unset, nothing is published. Kubernetes realizations refuse it; reach the Service with kubectl port-forward, or attach core/public-ingress.
engine_versionstringv0.8.7nonLibreChat container image tag. Selects the image tag of the platform container on both Recipes.
source_versionstring—nonVersion of the platform's own Helm chart the chart-backed Recipe pulls, defaulting to the version that Recipe was checked against. The workload-synthesis Recipe pulls no chart and reads nothing here.
storageobject—non—
external_base_urlstringnonExternally reachable base URL of the installation, as a browser sees it — the address the ingress tier publishes. The platform composes the links it emits from it and accepts browser requests against it. Left empty, the Kubernetes Recipes supply the installation's own in-cluster address instead, which suits an installation reached only in-cluster, and the Docker Compose Recipe http://localhost:<port>; where infra.published_port differs from infra.port there, set this to http://127.0.0.1:<published_port>.
computeobject—non—
overridesobject (open)—nonProvider-specific overrides. Accepts any nested structure. Each Recipe documents which keys it reads.

Contrat de sortie

Valeurs que ce composant publie vers les consommateurs via la topologie uses.

NomTypeValeur par défautObligatoireDescription
hoststring—ouiIn-cluster DNS name of the Service serving the chat interface and its API. The ingress tier points its upstream at this name.
portinteger—ouiTCP port that Service serves the interface and API on. It follows the port an operator set in the Release, so a consumer reads it rather than assuming a default.
protocolstring—nonURL scheme to reach `host` and `port` over, which a consumer composes its base URL on.
metrics_endpointstring—nonThe full address Prometheus scrapes this component's metrics at — scheme, host, port and path, `http://<host>:<port>/metrics`, on the port the interface is served on. Empty while values.metrics_enabled is false.
metrics_credential_requiredboolean—nonWhether a scrape of `metrics_endpoint` must present a credential: true while values.metrics_enabled is true, because the platform answers its metrics only to the secret attached to metrics_secret, which a scrape presents on its connection's `credential` slot; false while no address is published.

Sockets uses

Les sockets de dépendance par lesquels ce composant se connecte. La cible de chaque connexion câblée via un socket doit satisfaire son sélecteur d’étiquettes, et chaque connexion peut porter une attache dans les emplacements que le socket déclare. Un emplacement obligatoire doit être rempli par chaque connexion que l’Aggregate ou la Release câble via le socket.

model

Sélecteur
component.aggrega.ai/category in (ai-model, gateway), model.aggrega.ai/api
Injection
—

Emplacements que chaque connexion passant par ce socket peut remplir

EmplacementSélecteurObligatoireInjectionSensibleDescription
credentialfeature.aggrega.ai/kind=secretnonenvouiOptional credential the platform presents to the model this connection reaches. Unattached, the Recipe leaves the platform in its per-user mode, where each person enters their own key in the interface and no key is rendered at all.

backend

Sélecteur
component.aggrega.ai/category=datastore
Injection
—

Emplacements que chaque connexion passant par ce socket peut remplir

EmplacementSélecteurObligatoireInjectionSensibleDescription
credentialfeature.aggrega.ai/kind=secretnonenvouiOptional sign-in the platform presents to the document store this connection reaches, written as <user>:<password> with any reserved URL character in either part percent-encoded, since it becomes the user-information part of the connection string. Unattached, the connection string carries no user part.

Emplacements d’attache

Points de liaison nommés auxquels une Feature s’attache à ce composant. Le sélecteur d’un emplacement restreint la Feature qui peut le remplir, et sa liste inject indique comment le matériel de la Feature peut être livré. Un emplacement obligatoire doit être lié sur chaque instance, par une attache sur l’instance ou à une portée supérieure, et un emplacement sensible est déclaré pour traiter du matériel secret. Une Feature peut aussi s’attacher sous un nom qu’aucun emplacement ne déclare.

EmplacementSélecteurObligatoireInjectionSensibleDescription
session_secretfeature.aggrega.ai/kind=secretouienvouiRequired secret the platform signs its access tokens with. There is no default: the build halts when nothing fills this slot. Blanking the variable instead, so the platform generates one for itself, is a bootstrap convenience rather than a credential-rotation system: the generated value does not survive a restart and a second replica generates a different one, and each change invalidates every session then active.
session_refresh_secretfeature.aggrega.ai/kind=secretouienvouiRequired secret the platform signs its refresh tokens with — the tokens a browser presents to mint a new access token, so it carries the weight of the access-token secret. There is no default: the build halts when nothing fills this slot. Blanking the variable instead, so the platform generates one for itself, is a bootstrap convenience rather than a credential-rotation system: the generated value does not survive a restart and a second replica generates a different one, and each change makes every person sign in again.
credential_encryption_keyfeature.aggrega.ai/kind=secretouienvouiRequired key the platform encrypts the credentials it stores in the document store under, as 64 hexadecimal characters (a 32-byte key). It encrypts the credentials it supports whether or not the people using the installation ever save one of their own, so this slot is filled even where the model key is supplied centrally. There is no default: the build halts when nothing fills this slot. Blanking the variable instead, so the platform generates one for itself, is a bootstrap convenience rather than a credential-rotation system: the generated value does not survive a restart and a second replica generates a different one, and each change leaves everything already stored undecryptable.
credential_encryption_ivfeature.aggrega.ai/kind=secretouienvouiRequired initialization vector the platform pairs with the credential-encryption key when it encrypts what it stores, as 32 hexadecimal characters (16 bytes). A vector is not secret by cryptographic design, but one every reader of this catalog held would be the same in every Release, so there is no default: the build halts when nothing fills this slot. Blanking the variable instead, so the platform generates one for itself, is a bootstrap convenience rather than a credential-rotation system: the generated value does not survive a restart and a second replica generates a different one, and each change leaves everything already stored undecryptable.
metrics_secretfeature.aggrega.ai/kind=secretnonenvouiOptional secret the platform answers its Prometheus metrics to: a scrape presents it as a bearer token, on its connection's credential slot. It is delivered as METRICS_SECRET only while values.metrics_enabled is true, and is required then: the build halts when the value is true and nothing fills this slot. There is no default. While the value is false it is not delivered, and the platform answers its metrics route 401 to every caller.

Liste d’autorisation d’auto-injection

Aucun canal d’auto-injection déclaré.

Relations

Arêtes typées vers et depuis d’autres entités publiées. Une référence sans cible publiée s’affiche comme une arête rompue plutôt que comme un lien.

Utilise

Utilisé par

Référence du concept Component →

Le catalogue Aggrega — un index ouvert de composants, de fonctionnalités, de recettes et des plugins qui en assurent le rendu.

Conçu avec le système de design partagé d’Aggrega.

Confidentialité