Loading...
Skip to main content

Create a Corpus.

A Corpus consists of one or more Sources, where each Source defines a set of rules for crawling and indexing Documents.

Request Body required
    corpus object

    A queryable collection of documents. A corpus can be used to ground an LLM with factual content for a particular domain.

    owner object

    The the owner of this corpus.

    user object

    Represents a single Fixie User.

    email string

    The email address for this user.

    fullName string

    The name of the user. This is only used for display purposes.

    avatarUrl string

    The URL of the avatar image for this user.

    apiToken string

    The user's API token. This is only populated when a user accesses their own account information.

    team object

    Represents a Fixie Team.

    displayName string

    The display name for this team.

    description string

    The description for this team.

    avatarUrl string

    The URL of the avatar image for this team.

    members object[]

    The members of this Team.

  • Array [
  • user object

    The user that is a member of this Team.

    email string

    The email address for this user.

    fullName string

    The name of the user. This is only used for display purposes.

    avatarUrl string

    The URL of the avatar image for this user.

    apiToken string

    The user's API token. This is only populated when a user accesses their own account information.

    role object

    The user's role on the Team.

    isAdmin boolean

    True if the user is an admin for the team.

  • ]
  • displayName string

    The human-readable name for this corpus.

    stats object

    The current stats for this corpus.

    This field should not be populated in client requests.

    description string

    A human-readable description of this corpus.

    public boolean

    Whether this corpus is public. A public corpus is visible to other users of the Fixie platform, and can be queried by any agent. Only the corpus owner may modify its settings, however.

    jobCallbacks object[]

    Callbacks to execute when jobs reach particular states.

  • Array [
  • stateFilter enum[] required

    Possible values: [JOB_STATE_UNSPECIFIED, JOB_STATE_PENDING, JOB_STATE_RUNNING, JOB_STATE_COMPLETED, JOB_STATE_FAILED, JOB_STATE_CANCELLED]

    The callback will be triggered when a Job enters one of these states.

    webhook object

    A Webhook callback. When a job reaches the specified state, a POST request will be sent to the specified URL. The WebhookBody can be serialized as JSON or Protobuf depending on the Content-Type header and will be filtered based on the provided FieldMask. Also see fixie_api.corpus.v1.WebhookBody.

    url string required

    The URL to POST to.

    headers object

    The HTTP headers to send. Include a Content-Type header to explicitly choose how the Job is serialized. If no Content-Type is provided, we'll use JSON (and add the appropriate application/json header).

    property name* string
    bodyFieldMask field-mask

    A FieldMask to apply to the body before serializing it. If not provided, all fields will be included.

    doNotRetry boolean

    If set, the callback will only be attempted once even if the request fails. Otherwise failures will be retried a few times with backoff between them.

    email object

    An email callback. When a job reaches the specified state, an email will be sent to the specified recipients.

    to object[] required
  • Array [
  • name string
    email string required
  • ]
  • corpusStatusFilter enum[]

    Possible values: [CORPUS_STATUS_UNSPECIFIED, CORPUS_STATUS_EMPTY, CORPUS_STATUS_INITIALIZING, CORPUS_STATUS_READY, CORPUS_STATUS_UPDATING]

    The callback will only be sent if the corpus is in one of these states at the trigger time.

    rateLimit object

    Callbacks will be sent no more frequently than this limit. (They may be triggered more frequently, but won't be sent until quota is available again. Only sent callbacks count against the quota. They also only count once regardless of whether they end up being retried.)

    numRequests int32

    The number of events allowed per duration.

    window string

    Possible values: Value must match regular expression ^-?(?:0|[1-9][0-9]{0,11})(?:\.[0-9]{1,9})?s$

    The duration over which the events are allowed. This defines a rolling window.

  • ]
  • teamId string

    If specified, the corpus will be created in the specified team. Otherwise, the corpus will be owned by the user making the request.

Responses

OK


Schema
    corpus object

    A queryable collection of documents. A corpus can be used to ground an LLM with factual content for a particular domain.

    corpusId string

    The unique ID of this corpus.

    This field should not be populated in client requests.

    owner object

    The the owner of this corpus.

    user object

    Represents a single Fixie User.

    userId string

    The unique user ID for this user.

    This field should not be populated in client requests.

    email string

    The email address for this user.

    fullName string

    The name of the user. This is only used for display purposes.

    avatarUrl string

    The URL of the avatar image for this user.

    created date-time

    When this user was created.

    This field should not be populated in client requests.

    modified date-time

    When this user was last modified.

    This field should not be populated in client requests.

    apiToken string

    The user's API token. This is only populated when a user accesses their own account information.

    lastLogin date-time

    When this user last logged in.

    This field should not be populated in client requests.

    team object

    Represents a Fixie Team.

    teamId string

    The unique ID for this team.

    This field should not be populated in client requests.

    displayName string

    The display name for this team.

    description string

    The description for this team.

    avatarUrl string

    The URL of the avatar image for this team.

    members object[]

    The members of this Team.

  • Array [
  • teamId string

    The team ID.

    user object

    The user that is a member of this Team.

    userId string

    The unique user ID for this user.

    This field should not be populated in client requests.

    email string

    The email address for this user.

    fullName string

    The name of the user. This is only used for display purposes.

    avatarUrl string

    The URL of the avatar image for this user.

    created date-time

    When this user was created.

    This field should not be populated in client requests.

    modified date-time

    When this user was last modified.

    This field should not be populated in client requests.

    apiToken string

    The user's API token. This is only populated when a user accesses their own account information.

    lastLogin date-time

    When this user last logged in.

    This field should not be populated in client requests.

    role object

    The user's role on the Team.

    isAdmin boolean

    True if the user is an admin for the team.

    pending boolean

    Whether the membership is pending acceptance of an invitation by the user.

    created date-time

    When this membership was created.

    modified date-time

    When this membership was last modified.

  • ]
  • created date-time

    When this team was created.

    This field should not be populated in client requests.

    modified date-time

    When this team was last modified.

    This field should not be populated in client requests.

    displayName string

    The human-readable name for this corpus.

    created date-time

    When this corpus was created.

    This field should not be populated in client requests.

    modified date-time

    When this corpus was last modified.

    This field should not be populated in client requests.

    stats object

    The current stats for this corpus.

    This field should not be populated in client requests.

    status enum

    Possible values: [CORPUS_STATUS_UNSPECIFIED, CORPUS_STATUS_EMPTY, CORPUS_STATUS_INITIALIZING, CORPUS_STATUS_READY, CORPUS_STATUS_UPDATING]

    The current status of this corpus, indicating whether it is queriable.

    lastUpdated date-time

    The last time a job completed for any source in this corpus.

    numChunks string

    The total number of chunks in this corpus. Chunks are subsets of documents sized to respect context window limitations of LLMs.

    numDocs string

    The total number of documents in this corpus.

    numVectors string

    The total number of vectors in this corpus. Vectors are used for semantic search. Multiple vectors may correspond to a single chunk.

    description string

    A human-readable description of this corpus.

    public boolean

    Whether this corpus is public. A public corpus is visible to other users of the Fixie platform, and can be queried by any agent. Only the corpus owner may modify its settings, however.

    jobCallbacks object[]

    Callbacks to execute when jobs reach particular states.

  • Array [
  • stateFilter enum[] required

    Possible values: [JOB_STATE_UNSPECIFIED, JOB_STATE_PENDING, JOB_STATE_RUNNING, JOB_STATE_COMPLETED, JOB_STATE_FAILED, JOB_STATE_CANCELLED]

    The callback will be triggered when a Job enters one of these states.

    webhook object

    A Webhook callback. When a job reaches the specified state, a POST request will be sent to the specified URL. The WebhookBody can be serialized as JSON or Protobuf depending on the Content-Type header and will be filtered based on the provided FieldMask. Also see fixie_api.corpus.v1.WebhookBody.

    url string required

    The URL to POST to.

    headers object

    The HTTP headers to send. Include a Content-Type header to explicitly choose how the Job is serialized. If no Content-Type is provided, we'll use JSON (and add the appropriate application/json header).

    property name* string
    bodyFieldMask field-mask

    A FieldMask to apply to the body before serializing it. If not provided, all fields will be included.

    doNotRetry boolean

    If set, the callback will only be attempted once even if the request fails. Otherwise failures will be retried a few times with backoff between them.

    email object

    An email callback. When a job reaches the specified state, an email will be sent to the specified recipients.

    to object[] required
  • Array [
  • name string
    email string required
  • ]
  • corpusStatusFilter enum[]

    Possible values: [CORPUS_STATUS_UNSPECIFIED, CORPUS_STATUS_EMPTY, CORPUS_STATUS_INITIALIZING, CORPUS_STATUS_READY, CORPUS_STATUS_UPDATING]

    The callback will only be sent if the corpus is in one of these states at the trigger time.

    rateLimit object

    Callbacks will be sent no more frequently than this limit. (They may be triggered more frequently, but won't be sent until quota is available again. Only sent callbacks count against the quota. They also only count once regardless of whether they end up being retried.)

    numRequests int32

    The number of events allowed per duration.

    window string

    Possible values: Value must match regular expression ^-?(?:0|[1-9][0-9]{0,11})(?:\.[0-9]{1,9})?s$

    The duration over which the events are allowed. This defines a rolling window.

  • ]
Loading...