Advanced Analytica Advanced Analytica: IBOM
← All connectors
Ashby logo

Source connector

Ashby

Set up Ashby as a source and move its data into your chosen destination.

Version 2.1.4 certified alpha

Ashby

Use this page to set up the Ashby source connector and find reference information.

Before you start

You need:

  • An Ashby account
  • An Ashby API key with appropriate permissions for the streams you want to sync. See Ashby authentication docs for details about creating an API key.
  • A start date in YYYY-MM-DDTHH:MM:SSZ format. The connector only replicates applications, application feedback, and interview schedules created on or after this date. See setup guide to understand how the start date affects other streams.

Enable read permissions for the modules that match the streams you want to sync:

Ashby permission moduleStreams
Candidatesapplications, application_criteria_evaluations, application_feedback, application_history, candidates
Interviewsinterviews, interview_plans, interview_schedules, interview_stage_groups, interview_stages
Jobsjob_boards, job_postings, job_templates, jobs, openings
Hiring Processarchive_reasons, candidate_tags, custom_fields, feedback_form_definitions, interviewer_pools, source_tracking_links, sources, survey_form_definitions
Organization (always required)departments, locations, users — The connection check validates connectivity using the users stream, so you must enable this permission even if you only intend to sync streams from other modules. Without it, the check fails with a 403 missing_endpoint_permission error.
Offersoffers
Approvalsapprovals
Projects or Candidatesprojects (Ashby accepts either permission)

Ashby API keys can’t read confidential jobs and projects by default. To sync them, enable Allow access to confidential jobs and projects? on the API key.

To read private candidate fields outside offers, enable Allow access to non-offer private fields? on the API key. Module read permissions don’t grant access to these fields.

:::note The AI Application Review feature must be enabled for your Ashby account to use the application_criteria_evaluations stream. Without this feature, the stream returns empty results. :::

Set up Ashby

  1. Log in to your Ashby account.
  2. Generate an API key by following Ashby authentication guide. Grant read permissions for the modules listed in the prerequisites. You must enable at least the Organization read permission, which the connection check requires. Enable read permissions for any other modules whose streams you want to sync.
  3. In Airbyte, create a new Ashby source.
  4. Enter your API key.
  5. Enter a Start date in YYYY-MM-DDTHH:MM:SSZ format. The connector sends this date as the createdAfter filter for the application_feedback, applications, and interview_schedules streams. Records created before this date aren’t replicated. The date also limits application_criteria_evaluations and application_history. These streams use the same filtered application list to identify applications requiring child records. All other streams ignore the start date, including candidates. Their initial data loads aren’t limited by this date.

Supported sync modes

FeatureSupported
Full RefreshYes
Incremental - AppendYes, for every stream except application_criteria_evaluations, archive_reasons, interview_stage_groups, interview_stages, job_boards, job_postings, and sources
Incremental - Append + DedupedYes, for every stream except application_criteria_evaluations, archive_reasons, interview_stage_groups, interview_stages, job_boards, job_postings, and sources

For existing connections, refresh the source schema after upgrading to make new sync modes available. Then select Incremental for each stream you want to sync incrementally.

Sync-token streams

From version 2.1.0, all incremental streams except applications and application_history use Ashby’s sync tokens. The first incremental sync reads every available record, subject to the start date where applicable. After reading the final page, the connector saves a token. Later syncs use this token to request only new or modified records. Airbyte manages these tokens automatically, so you don’t need to enter one in the source configuration.

If Ashby rejects a stored token, the stream fails but keeps its state. For sync_token_expired or sync_token_invalid, refresh the affected stream to clear its state and read it fully again.

:::warning Ashby limits each sync-token incremental sync to 100 pages. This limit represents about 10,000 records changed since the stream’s previous sync. A larger change set can occur after a bulk data migration. It fails the stream with incremental_sync_too_large. Syncing again with the same state produces the same failure. To recover, switch the stream to Full Refresh. Alternatively, refresh the stream and sync the connection more often, at least daily. Ashby’s sync tokens expire after 14 days, so sync incremental streams at least weekly. :::

Applications and application history

From version 1.5.0, applications and application_history support incremental sync using the application’s updatedAt timestamp. These streams don’t use sync tokens, so token expiry and page-depth limits don’t apply.

Ashby’s application.list doesn’t filter on updatedAt. An incremental sync of applications therefore still reads every application from Ashby. It only emits applications updated within one day before the previous sync’s latest updatedAt. This one-day lookback prevents missed updates because Ashby returns applications in creation order. Applications updated during this period are emitted again during the next sync. Use Incremental | Append + Deduped, keyed on id, to keep one row per application. Every incremental sync still reads the full application list. The time savings come from application_history, described below. Refresh the source schema to see the application_updated_at column on application_history.

Supported streams

This source syncs these streams:

The application_criteria_evaluations stream is a substream of applications. The connector only requests evaluations for applications whose current interview stage has the type PreInterviewScreen. Their status must also be neither Archived nor Hired. The stream therefore doesn’t cover every application in your account. Each record includes an application_id field copied from its parent application. Use this field to join evaluations to applications. From version 2.0.0, the stream’s primary key is application_id and id. Earlier versions had no primary key. From version 1.4.0, the connector pages through every evaluation for each application. Earlier versions synced only the first page. Like application_history, this stream makes at least one request per application. The connector therefore caps application.listCriteriaEvaluations at 100 requests per minute. It skips applications when Ashby returns application_not_found.

The application_history stream is a substream of applications. Each record represents an interview stage entered by an application. It includes the enteredStageAt and leftStageAt timestamps, which no other Ashby endpoint provides. Join application_history.application_id to applications.id and application_history.stageId to interview_stages.id. The connector also copies the parent application’s status, creation timestamp, and update timestamp into each record. These fields are application_status, application_created_at, and application_updated_at, alongside application_id. You can therefore analyse stage timing without joining back to applications. The stream’s primary key is id. Deduplicating destinations therefore key history events instead of appending a copy during every sync.

The parent application list uses the same createdAfter filter as the applications stream. A start date after your oldest application returns partial history, not an error.

The connector requests history for one application at a time. application.listHistory accepts neither a date filter nor a syncToken. The connector caps this endpoint at 100 requests per minute, setting a minimum duration for full syncs. For example, 10,000 applications require at least 100 minutes. Applications with over 100 history records require extra pagination requests.

In incremental mode, the first sync requests history for every application. Later syncs only request history for applications updated within one day before the previous sync’s latest updatedAt. This lookback prevents missing applications updated during the previous sync. Testing showed that Ashby updated updatedAt when an application moved to a new stage. New stage entries are therefore included. Ashby doesn’t update updatedAt for every change. Examples include history entries edited or deleted through application.updateHistory and some application changes. If these changes matter, run a periodic Refresh and remove records. Each changed application emits its full history again. Use Incremental | Append + Deduped, keyed on id, to keep one row per history event. Incremental | Append adds another history copy each time. Each incremental sync still reads the full application list to find changed applications. In full refresh mode, every sync reads the history of every selected application again. Sync it through a separate connection with an infrequent schedule.

Ashby can return an application_not_found error when an application is deleted or your API key can’t access it. The connector then skips that application’s history, logs the Ashby request ID, and continues. It retries HTTP 429 and 5xx responses. Any other error fails the sync.

The application_feedback stream returns submitted interview scorecards, with one record for each feedback form submission. Each record includes the submittedValues entered by an interviewer and the formDefinition active when they submitted the form. Join applicationId to applications.id, interviewId to interviews.id, and feedbackFormDefinitionId to feedback_form_definitions.id. submittedValues is a free-form object keyed by each field’s path. The connector therefore doesn’t declare its keys. For select fields, it stores the option value, such as hire, instead of the display label, such as Hire. To find labels, map each value through formDefinition.sections[].fields[].field.selectableValues on the same record. Don’t use the current feedback_form_definitions stream because form definitions can change after feedback is submitted. Ashby added the creditedToUser field to its API on 2026-07-21. This field may be null on older records. This endpoint requires the Candidates read permission.

The interview_stages stream is a substream of Ashby’s interview plans. Ashby’s interviewStage.list endpoint returns stages for one interview plan at a time. From version 1.4.0, the connector lists every interview plan, including archived plans, then requests each plan’s stages. Versions before 1.4.0 didn’t send an interview plan ID, so they synced no stages. Each record includes interviewPlanId and an interview_plan_is_archived flag copied from its parent plan. Join interviewPlanId to interview_plans.id and interviewStageGroupId to interview_stage_groups.id. Ashby’s interviewPlan.list doesn’t return draft plans. It only returns job-specific plans when your API key can read the associated job. Enable the Jobs read permission if you need job-specific plans in interview_plans or their stages in interview_stages.

The interviews stream returns interview definitions, such as a technical phone screen, configured in your Ashby account. It doesn’t return scheduled interviews. Each record includes the definition’s title, externalTitle, instructions, feedback settings, and feedbackFormDefinitionId. For scheduled interviews, including their times and interviewers, use interview_schedules.

From version 1.2.0, the connector sends includeNonSharedInterviews: true. Definitions belonging to one job therefore sync alongside shared definitions. Use jobId to distinguish them. It contains the related job for job-specific definitions and is null for shared definitions. You can schedule shared definitions against any job. From version 1.4.0, the connector also sends includeArchived: true, so archived definitions sync. Use isArchived to exclude them.

Version 1.2.0 also declared the fields returned by interview.list, which the schema previously omitted. Refresh the source schema and enable the new columns in your connection to replicate them. Version 2.0.0 removed twelve scheduled-interview fields because Ashby’s interview.list endpoint never returns them. These fields were applicationId, interviewScheduleId, interviewStageId, status, createdAt, updatedAt, cancelledAt, startTime, endTime, feedbackLink, interviewerUserIds, and meetingLink.

The openings stream returns headcount openings. Each record’s latestVersion object contains the opening’s current details, including jobIds and locationIds. These fields join to jobs.id and locations.id.

The approvals stream returns approval processes for offers, jobs, and openings. The connector doesn’t filter by entity, so every approval in your account syncs. Use entityType to identify the record type referenced by entityId. Then join entityId to offers.id, jobs.id, or openings.id.

Archived and deactivated records

From version 1.4.0, the connector asks Ashby to include archived records from several streams. These streams are archive_reasons, candidate_tags, custom_fields, departments, feedback_form_definitions, interviews, locations, and sources. It also includes deactivated users from users. Earlier versions only synced active records from these streams. Your first sync after upgrading may therefore return more rows. To include only active records, filter on isArchived. For users, filter on isEnabled.

Streams added in version 1.6.0 also include inactive records. interview_plans includes archived plans. interviewer_pools includes archived pools and archived training stages. Filter on isArchived to exclude these records. source_tracking_links includes disabled links, which you can exclude by filtering on enabled.

Among the streams added in version 1.7.0, job_templates returns active and inactive templates. Filter on status to include only active templates. Ashby’s jobBoard.list only returns enabled job boards, so job_boards doesn’t include disabled job boards.

Performance

Ashby documents a rate limit of 1,000 requests per minute per API key. The connector reads up to two streams simultaneously. An API key shared with other integrations has less available capacity. To protect this capacity, the connector limits application.listHistory and application.listCriteriaEvaluations to 100 requests per minute each. It calls each of these endpoints at least once per application. The connector applies no request budget to other endpoints. After an HTTP 429 response, it waits for the Retry-After interval and retries. If the header is missing, it applies exponential backoff before retrying.

Troubleshooting

Ashby reports most errors as HTTP 200 responses with success: false in the body. From version 1.4.0, the connector fails the sync after these responses. The error includes Ashby’s error code, message, and request ID. Earlier versions could finish without records or an error report.

  • HTTP 401 Unauthorized: Ashby rejected the API key. Check that the key in the source settings is correct and active. Alternatively, generate a new key in Ashby.
  • HTTP 403 Forbidden: The API key may be deactivated or lack access. Generate a new key in Ashby.
  • incremental_sync_too_large: About 10,000 records have changed since the stream’s previous incremental sync. Switch the stream to Full Refresh. Alternatively, refresh the stream and sync the connection more often, at least daily.
  • sync_token_expired or sync_token_invalid: The stored sync token is over 14 days old or can’t be read. Refresh the stream to read it fully again.
  • missing_endpoint_permission: The API key lacks a read permission required by a selected stream. The error message identifies the missing permission. Enable it on the key by using permission table, or deselect the stream.

The connector retries HTTP 429 and 5xx responses before it fails.

IP allow list

If you use Airbyte Cloud and restrict access to specific IPs, add the Airbyte Cloud IP addresses to your allow list.

Upgrade to 2.0.0

Version 2.0.0 corrects four stream schemas to match the Ashby API. applications.archiveReason becomes an object, and job_postings.publishedDate becomes a date. application_criteria_evaluations gains a primary key comprising application_id and id. Fields Ashby never returns are removed from interviews and application_criteria_evaluations. Refresh the source schema after upgrading. Because applications syncs incrementally, existing rows keep their old archiveReason value until the application changes in Ashby. Refresh the applications stream to rewrite them using the new shape. The destination may then fail to change a column’s type. Examples include a Snowflake invalid type ... for parameter 'TO_OBJECT' error or an S3 Data Lake or Iceberg schema-evolution error. If this happens, drop the affected tables and refresh those streams. See Ashby migration guide for every change, required downstream query update, and upgrade step.

Upgrade to 1.0.0

Version 1.0.0 declares element schemas for array columns that the connector previously left untyped. These columns change type in data-lake destinations such as S3 Data Lake and Iceberg. Syncs can therefore fail with a schema evolution error. Refresh the affected streams first. Only drop and recreate the affected destination tables if a sync still fails. See Ashby migration guide for every affected column and the upgrade steps.

Changelog

Expand to review
VersionDatePull RequestSubject
2.1.42026-11-1091399Update dependencies
2.1.32026-11-0390480Update dependencies
2.1.22026-10-3090220Promote to certified
2.1.12026-10-2990197Show a date picker for the start date field.
2.1.02026-10-2989675Added incremental sync using Ashby sync tokens for the 20 streams whose endpoints support them.
2.0.12026-10-2789792Update dependencies
2.0.02026-10-2387726Breaking: correct the applications, job_postings, application_criteria_evaluations, and interviews schemas to match the Ashby API. See the migration guide.
1.7.32026-10-2089142Update dependencies
1.7.22026-10-1388502Update dependencies
1.7.12026-10-0687745Update dependencies
1.7.02026-10-0587590Add openings, job templates, job boards, approvals, and projects streams
1.6.02026-10-0587587Add interview plans, interview stage groups, interviewer pools, survey form definitions, and source tracking links streams
1.5.02026-10-0287597Add incremental sync to applications and application_history, so incremental syncs request history only for applications updated since the previous sync, with a 1-day lookback
1.4.02026-09-3087051Fail syncs on Ashby success: false errors, sync interview_stages per interview plan, paginate application_criteria_evaluations, and include archived and deactivated records in lookup streams
1.3.42026-09-2987080Update dependencies
1.3.32026-09-2286515Update dependencies
1.3.22026-09-2186501Add mock-server tests and sandbox acceptance-test secret
1.3.12026-09-1585974Update dependencies
1.3.02026-09-0985755Add application_feedback stream; send the API key as the Basic auth username with a blank password per Ashby docs
1.2.12026-09-0885402Update dependencies
1.2.02026-08-2985183Declare the interview definition fields interview.list actually returns on the interviews stream, and request non-shared (job-specific) interviews
1.1.02026-08-2584392Add application history stream
1.0.12026-08-2584405Send the pagination page size using Ashby’s documented limit field instead of the undocumented per_page field
1.0.02026-08-1884274Breaking: declare documented API fields across stream schemas, including element schemas for previously untyped array columns. Data-lake users must refresh the affected streams, then recreate the affected tables if a sync still fails. See the migration guide.
0.3.92026-08-1878554Update dependencies
0.3.82026-08-1184215Promoted release candidate to GA
0.3.8-rc.52026-08-1184214Revert concurrency support
0.3.8-rc.42026-08-1183816Add missing application, candidate, and source fields to the declared schemas, and remove duplicated unreferenced manifest blocks.
0.3.8-rc.32026-05-2678434Decrease default concurrency to 2 and add explicit worker count plus API request budget for the next rollout.
0.3.8-rc.22026-05-2178307Decrease default concurrency to 3 after Phase 1 rollout monitoring found source-read regressions and a 429 retry warning.
0.3.8-rc.12026-05-1877048Add concurrency support with default_concurrency=4 for concurrent stream reads
0.3.72026-04-2877144Update dependencies
0.3.62026-04-2176510Update dependencies
0.3.52026-03-3175881Update dependencies
0.3.42026-03-2475325Update dependencies
0.3.32026-03-1074490Update dependencies
0.3.22026-02-2473805Update dependencies
0.3.12026-02-1760692Update dependencies
0.3.02026-02-1373244Add interviews, interview_stages, and application_criteria_evaluations streams
0.2.232025-05-1059853Update dependencies
0.2.222025-05-0359322Update dependencies
0.2.212025-04-2658746Update dependencies
0.2.202025-04-1958271Update dependencies
0.2.192025-04-1257150Update dependencies
0.2.182025-03-2956594Update dependencies
0.2.172025-03-2256140Update dependencies
0.2.162025-03-0855387Update dependencies
0.2.152025-03-0154888Update dependencies
0.2.142025-02-2254234Update dependencies
0.2.132025-02-1553874Update dependencies
0.2.122025-02-0853407Update dependencies
0.2.112025-02-0152893Update dependencies
0.2.102025-01-2552162Update dependencies
0.2.92025-01-1851710Update dependencies
0.2.82025-01-1151292Update dependencies
0.2.72024-12-2850493Update dependencies
0.2.62024-12-2150207Update dependencies
0.2.52024-12-1449572Update dependencies
0.2.42024-12-1249014Update dependencies
0.2.32024-11-0448196Update dependencies
0.2.22024-10-2947729Update dependencies
0.2.12024-10-2847616Update dependencies
0.2.02024-08-1944420Refactor connector to manifest-only format
0.1.162024-08-1744288Update dependencies
0.1.152024-08-1243780Update dependencies
0.1.142024-08-1043491Update dependencies
0.1.132024-08-0343080Update dependencies
0.1.122024-07-2742658Update dependencies
0.1.112024-07-2042220Update dependencies
0.1.102024-07-1742028Fix typo in application stream
0.1.92024-07-1341818Update dependencies
0.1.82024-07-1041379Update dependencies
0.1.72024-07-0941271Update dependencies
0.1.62024-07-0640971Update dependencies
0.1.52024-06-2540469Update dependencies
0.1.42024-06-2240107Update dependencies
0.1.32024-06-0639159[autopull] Upgrade base image to v1.2.2
0.1.22024-05-2838666Make connector compatible with Builder
0.1.12024-05-2038421[autopull] base image + poetry + up_to_date
0.1.02022-10-2218334Add Ashby Source Connector
Advanced Analytica

Most businesses are sitting on information they can’t use at speed. The people who hold it are busy, and when they leave, the knowledge goes with them. The documents are static, and the processes around them weren’t designed for AI.

Advanced Analytica works with these businesses to turn that information into structured data assets. Once they're integrated, AI systems can govern it, and the business can use it. We bring in the strategy, the specification and the AI skills to rework how the business operates.

MOVE FAST. STAY SAFE.