Greenhouse
Set up the Greenhouse source connector and find its reference information on this page.
Prerequisites
The connector reads the Greenhouse Harvest v3 API, which supports two authentication methods. You need one of these authentication methods:
- OAuth (Airbyte Cloud only): a Greenhouse Site Admin must approve the consent flow. Airbyte provides the client ID and client secret.
- Client Credentials: provide the client ID and client secret for a Harvest V3 (OAuth) custom integration credential created in Greenhouse. You can also provide a Site Admin’s user ID.
Whichever method you choose, the token needs the Harvest v3 scopes for each stream you sync. The Start date and Number of concurrent threads settings are optional.
Setup
Choose an authentication method and prepare it in Greenhouse. Then create the source in Airbyte.
Set up Greenhouse
Choose the method that matches your Airbyte deployment and credential source.
OAuth
OAuth (Authorization Code) authenticates through Airbyte’s registered Greenhouse partner application. You don’t create, request or register your own Greenhouse OAuth application. Greenhouse issues partner credentials only to integration partners, not Greenhouse customers. Airbyte provides the client ID and client secret during the consent flow. This method is available only in Airbyte Cloud because it depends on Airbyte’s partner credentials.
To use OAuth, you need:
- An Airbyte Cloud workspace.
- A Greenhouse Site Admin who can approve the consent flow.
Client Credentials
Client Credentials uses a Harvest V3 (OAuth) custom integration credential that you create in Greenhouse. Use it with self-managed Airbyte. You can also use it in Airbyte Cloud instead of Airbyte’s partner application.
To use Client Credentials, you need:
- Access Configure > Dev Center > API Credential Management in Greenhouse. Click Create new API credentials, select Harvest V3 (OAuth) and save the credential. Then grant the scopes listed below. Greenhouse provides a client ID and client secret. Copy both into the source.
- You can provide the numeric Greenhouse user ID of a Site Admin. Greenhouse creates an integration service user for every custom credential. By default, the connector sends requests as this user, which can read every Harvest v3 list endpoint. Enter a user ID as the Site Admin user ID only to attribute requests to a specific person. Greenhouse denies list endpoints to users who aren’t Site Admins.
Scopes
Both methods need the same scopes. For OAuth, approve every scope requested during the consent flow. For Client Credentials, grant these scopes to the Greenhouse credential:
harvest:application_stages:listharvest:applications:listharvest:applied_candidate_tags:listharvest:approval_flows:listharvest:approver_groups:listharvest:approvers:listharvest:attachments:listharvest:candidate_attribute_types:listharvest:candidate_educations:listharvest:candidate_employments:listharvest:candidate_tags:listharvest:candidates:listharvest:close_reasons:listharvest:custom_field_options:listharvest:custom_fields:listharvest:demographic_answer_options:listharvest:demographic_answers:listharvest:demographic_question_sets:listharvest:demographic_questions:listharvest:departments:listharvest:eeoc:listharvest:email_templates:listharvest:interview_kits:listharvest:interviewer_tags:listharvest:interviewers:listharvest:interviews:listharvest:job_candidate_attributes:listharvest:job_hiring_managers:listharvest:job_interview_stages:listharvest:job_interviews:listharvest:job_notes:listharvest:job_owners:listharvest:job_post_locations:listharvest:job_posts:listharvest:jobs:listharvest:notes:listharvest:offers:listharvest:offices:listharvest:openings:listharvest:prospect_details:listharvest:prospect_pool_stages:listharvest:prospect_pools:listharvest:referrers:listharvest:rejection_details:listharvest:rejection_reasons:listharvest:scorecard_candidate_attributes:listharvest:scorecard_question_answer_options:listharvest:scorecard_question_answers:listharvest:scorecard_question_options:listharvest:scorecard_questions:listharvest:scorecards:listharvest:sources:listharvest:user_emails:listharvest:user_job_permissions:listharvest:user_roles:listharvest:users:list
Harvest v3 rejects list endpoint requests from users who aren’t Site Admins or custom integration service users. The connector then stops the sync with a configuration error. A missing scope causes the same failure for dependent streams. Grant every listed scope unless you plan to disable its corresponding streams. Always grant harvest:users:list. The connection check reads the users stream, so setup fails without this scope, even if you never sync it.
With OAuth, Greenhouse links scopes to the refresh token issued when you approved consent. Versions 1.2.0, 1.4.0, and 1.5.0 each added streams requiring new scopes. Tokens created before one of these versions don’t include its added scopes. Enabling streams added in 1.2.0, streams added in 1.4.0, or streams added in 1.5.0 then causes a 403 error. Open the source settings, click Authenticate and approve consent again. Existing synced streams keep working without re-authentication. The consent flow requests scopes for the source’s current version. Re-authenticate after an upgrade, not before. Connections that propagate all field and stream changes enable new streams automatically. Syncs then fail with a 403 until you re-authenticate or disable those streams. With Client Credentials, grant the new scopes to the Greenhouse credential instead.
Set up the connector in Airbyte
- Log into your Airbyte Cloud account or open your self-managed Airbyte instance.
- Click Sources, then click + New source.
- On the Set up the source page, select Greenhouse from the Source type dropdown.
- Enter a name for the Greenhouse connector.
- Under Authentication, select OAuth or Client Credentials.
- For OAuth (Airbyte Cloud only), click Authenticate. Sign in to Greenhouse as a Site Admin and approve the requested scopes. Airbyte enters its partner application’s client ID and client secret, then stores the resulting refresh token. You don’t enter any Greenhouse credentials.
- For Client Credentials, enter the custom integration credential’s Client ID and Client secret. Leave Site Admin user ID blank to send requests as the credential’s integration service user. Enter it only to send requests as a specific Site Admin.
- You can enter a Start date in UTC using the format
YYYY-MM-DDTHH:MM:SSZ. Records updated before this date won’t be replicated. If you omit it, the connector replicates all history. - You can change Number of concurrent threads. The connector uses 2 threads by default and accepts 1 to 8. All threads share one Greenhouse rate limit. Increase this only if your Greenhouse account can handle more API traffic. Reduce it to 1 if syncs fail with rate-limit errors.
- Click Set up source.
:::warning If you use OAuth, sync at least once every 14 days. Greenhouse refresh tokens expire after 14 days of non-use and rotate on every refresh. If a connection pauses, stops or fails for over 14 days, repeat consent from the source settings. See Sync fails with a configuration error asking you to re-authenticate for the resulting error. Client Credentials has no refresh token, so this doesn’t apply. :::
Supported sync modes
The Greenhouse source connector supports these sync modes:
Start date filters the Greenhouse request, not the sync. Every incremental stream listed below sends updated_at=gte|<start date> with each sync. This includes streams configured as Full refresh. A full refresh therefore returns only records updated on or after your start date. Leave Start date empty to replicate all history.
Incremental streams reread a one-hour window before the saved cursor during each sync. This captures records updated shortly before the previous sync finished. Incremental - Append can therefore create duplicate destination records. Use Incremental - Append + Deduped, where supported, to keep only each record’s latest version.
Supported streams
New connections enable 10 streams by default: applications, candidates, jobs, job_posts, offers, interviews, scorecards, users, departments, and offices. Enable other streams from the connection’s stream list.
The table shows each Airbyte stream name and its Harvest v3 endpoint. Start date applies to every stream marked incremental, regardless of its selected sync mode. It doesn’t apply to other streams. Only the four demographics question and answer-option streams use full refresh. Harvest v3 provides no date filter on /v3/demographic_questions or /v3/demographic_answer_options, so these streams always read every returned record. Their two child streams retrieve parent IDs across your full Greenhouse history. Their coverage therefore doesn’t depend on Start date.
| Stream | Sync mode | Notes |
|---|---|---|
activity_feed | Incremental (updated_at) | Notes across all candidates |
application_stages | Incremental (updated_at) | Stage history per application, with time in stage. One row per application per stage entered |
applications | Incremental (updated_at) | |
applied_candidate_tags | Incremental (updated_at) | Candidate-to-tag assignments; tags on its own is only the dictionary |
approvals | Incremental (updated_at) | |
approver_groups | Incremental (updated_at) | The steps of an approval chain, per approval flow |
approvers | Incremental (updated_at) | Individual approvers and their status within a group |
attachments | Incremental (updated_at) | One file per application. url is a download link Greenhouse expires after seven days |
candidate_attribute_types | Incremental (updated_at) | The scorecard attribute dictionary, defined per job |
candidate_educations | Incremental (updated_at) | School, degree and discipline history per candidate |
candidate_employments | Incremental (updated_at) | Employer and title history per candidate |
candidates | Incremental (updated_at) | |
close_reasons | Incremental (updated_at) | |
custom_field_options | Incremental (updated_at) | Every custom field option in the account |
custom_fields | Incremental (updated_at) | |
degrees | Incremental (updated_at) | Custom field options for the degree field |
demographics_answer_options | Full refresh | |
demographics_answers | Incremental (updated_at) | |
demographics_answers_answer_options | Full refresh | Answer options for each question in demographics_questions |
demographics_question_sets | Incremental (updated_at) | |
demographics_question_sets_questions | Full refresh | Questions in each set in demographics_question_sets |
demographics_questions | Full refresh | |
departments | Incremental (updated_at) | |
disciplines | Incremental (updated_at) | Custom field options for the discipline field |
eeoc | Incremental (submitted_at) | |
email_templates | Incremental (updated_at) | |
interview_kits | Incremental (updated_at) | Kit content per job interview. exercises is the live value and may contain HTML |
interviewer_tags | Incremental (updated_at) | The interviewer tag dictionary |
interviewers | Incremental (updated_at) | One row per panel member on a scheduled interview |
interviews | Incremental (updated_at) | |
job_candidate_attributes | Incremental (updated_at) | The scorecard attributes rated on each job. scorecard_candidate_attributes.job_candidate_attribute_id references this stream |
job_hiring_managers | Incremental (updated_at) | One row per hiring manager on a job |
job_interviews | Incremental (updated_at) | The interview plan for each job stage |
job_notes | Incremental (updated_at) | Free-text notes on jobs, separate from the candidate notes in activity_feed |
job_owners | Incremental (updated_at) | Recruiter, sourcer and coordinator per job, with a responsible flag |
job_post_locations | Incremental (updated_at) | One row per location attached to a job post, published or not, resolved to an office, a custom location or free text |
job_posts | Incremental (updated_at) | Includes deleted posts |
job_stages | Incremental (updated_at) | |
jobs | Incremental (updated_at) | |
jobs_openings | Incremental (updated_at) | Openings across all jobs |
offers | Incremental (updated_at) | |
offices | Incremental (updated_at) | |
prospect_details | Incremental (updated_at) | Pool, stage and owner per prospect application |
prospect_pool_stages | Incremental (updated_at) | The stage dictionary for each prospect pool |
prospect_pools | Incremental (updated_at) | |
referrers | Incremental (updated_at) | Referral attribution per candidate |
rejection_details | Incremental (updated_at) | Reason, rejecter and note per rejected application |
rejection_reasons | Incremental (updated_at) | Includes the reasons Greenhouse ships with |
schools | Incremental (updated_at) | Custom field options for the school_name field |
scorecard_candidate_attributes | Incremental (updated_at) | One row per rated attribute on each scorecard |
scorecard_question_answer_options | Incremental (updated_at) | The options picked on each single_select or multi_select answer |
scorecard_question_answers | Incremental (updated_at) | Interviewers’ answers to scorecard questions. One row per question on each scorecard, including questions left blank |
scorecard_question_options | Incremental (updated_at) | The choices for single_select and multi_select scorecard questions |
scorecard_questions | Incremental (updated_at) | The question dictionary behind scorecards, defined per interview kit |
scorecards | Incremental (updated_at) | |
sources | Incremental (updated_at) | |
tags | Incremental (updated_at) | Candidate tags |
user_emails | Incremental (updated_at) | Every email address per user; users.primary_email is only one of them |
user_permissions | Incremental (updated_at) | Job permissions across all users |
user_roles | Incremental (updated_at) | |
users | Incremental (updated_at) | Includes integration service users |
Streams added in 1.5.0
Version 1.5.0 added these 5 streams, all incremental on updated_at. Together with scorecard_questions, the scorecard streams include each scorecard’s content, not just its overall recommendation. job_candidate_attributes names the attributes rated by scorecard_candidate_attributes:
job_candidate_attributes, job_post_locations, scorecard_question_answer_options, scorecard_question_answers, scorecard_question_options
All 5 are disabled by default on new and existing connections. Each needs its own Harvest v3 scope. If you authorised an OAuth source before 1.5.0, repeat consent after upgrading and before enabling these streams. Connections propagating all field and stream changes enable all five when 1.5.0 arrives. Syncs then fail with a 403 configuration error until you repeat consent or disable the streams. To prevent this, select propagation of field changes only before upgrading. With Client Credentials, grant harvest:job_candidate_attributes:list, harvest:job_post_locations:list, harvest:scorecard_question_answer_options:list, harvest:scorecard_question_answers:list, and harvest:scorecard_question_options:list to the Greenhouse credential. See Scopes.
Streams added in 1.4.0
Version 1.4.0 added these 2 streams, both incremental on updated_at:
candidate_attribute_types, job_notes
Both streams are disabled by default on new and existing connections. However, Airbyte enables them during upgrades when the connection propagates all field and stream changes. Each stream needs its own Harvest v3 scope. For OAuth sources authorised before 1.4.0, repeat consent before enabling either stream. With Client Credentials, grant harvest:candidate_attribute_types:list and harvest:job_notes:list to the Greenhouse credential. See Scopes.
Streams added in 1.2.0
Version 1.2.0 added these 20 streams, all incremental on updated_at. They contain details embedded in Harvest v1 parent records but served through separate Harvest v3 endpoints:
application_stages, applied_candidate_tags, approver_groups, approvers, attachments, candidate_educations, candidate_employments, interview_kits, interviewer_tags, interviewers, job_hiring_managers, job_interviews, job_owners, prospect_details, prospect_pool_stages, referrers, rejection_details, scorecard_candidate_attributes, scorecard_questions, user_emails
All 20 streams are disabled by default on new and existing connections. However, Airbyte enables them during upgrades when the connection propagates all field and stream changes. Each stream needs its own Harvest v3 scope. For OAuth sources authorised before 1.2.0, repeat consent before enabling any stream. With Client Credentials, grant the corresponding scopes to the Greenhouse credential. See Scopes.
Streams made incremental in 1.1.0
These 18 streams used full refresh before 1.1.0. They are now incremental on updated_at:
activity_feed, approvals, close_reasons, custom_field_options, custom_fields, degrees, demographics_question_sets, departments, disciplines, jobs_openings, offices, prospect_pools, rejection_reasons, schools, sources, tags, user_permissions, user_roles
This isn’t a breaking change. Schemas, primary keys and existing sync modes remain unchanged. There was no stream state to migrate, so you don’t need to act. Connections continue syncing.
One behaviour changes: Start date now applies to these 18 streams in every sync mode. Earlier versions sent no date filter and read your full Greenhouse history, regardless of Start date. With a start date, these streams now return only records updated on or after it. This applies to full refresh and incremental syncs because Greenhouse filters the request. On a Full refresh | Overwrite connection, this also removes older rows from the destination table. Each sync replaces the table with the records it reads. Existing rows remain on append and append + deduped connections, but stop refreshing. Clear Start date to continue reading full history, then refresh the streams.
Migrate from Harvest v1
Version 1.0.0 moves the 33 streams from 0.8.1 from Harvest v1 to Harvest v3. It also adds the new custom_field_options stream, bringing the total to 34. Greenhouse has scheduled Harvest v1 and v2 support to end together on 2026-08-31. Version 1.0.0 also replaces API-key authentication with OAuth Authorization Code authentication. It introduces an optional Start date that preserves full-history behaviour when omitted. Version 1.3.0 added Client Credentials for self-managed deployments that can’t use OAuth consent. These deployments can authenticate with a credential you create in Greenhouse. Create a new connection on 1.0.0 instead of refreshing the existing one. See the upgrade paths before upgrading.
IP allow list
If you use Airbyte Cloud and restrict access to specific IPs, add the Airbyte Cloud IP addresses to your allow list.
Performance
Greenhouse rate limits Harvest v3 in fixed 30-second windows. Each response shows the remaining allowance in X-RateLimit-Remaining. It shows the current window’s reset time in X-RateLimit-Reset. Greenhouse doesn’t publish a fixed Harvest v3 request ceiling. It also gives custom and partner integrations different allowances. The connector therefore limits itself to a conservative 50 requests per window and tracks these headers. When Greenhouse returns 429, the connector waits for the Retry-After interval. Every thread uses the same window. Syncing many streams with a high Number of concurrent threads commonly causes rate-limit errors. Lower this value before creating an issue about rate limits.
The connector requests 500 records per page, the Harvest v3 maximum. It then follows Greenhouse’s cursor links. Large accounts can still require many requests for each stream.
application_stages, scorecard_candidate_attributes, and scorecard_question_answers are high-volume streams. application_stages contains one row for each application and stage entered. scorecard_candidate_attributes contains one row for every rated attribute on each scorecard. scorecard_question_answers contains one row for every question on each scorecard. Each stream contains several times more rows than its parent. They add many requests and consume the shared rate-limit window. Leave them disabled unless you need this detail.
Limitations and troubleshooting
Limitations
eeocreplicates onsubmitted_at. Correcting an EEOC response after submission doesn’t changesubmitted_at, so incremental syncs don’t reread it. Refresh the stream to replicate corrections.custom_field_optionsreads every custom field option in your account. It is therefore a superset ofdegrees,disciplines, andschools. Those three streams read the same Greenhouse endpoint, filtered to one field key. They also share the same primary keys. Enabling all four writes identical option rows to four destination tables. Enable only the streams you need.usersincludes integration service users, which Greenhouse hides by default. Service accounts have no email address, soprimary_emailis empty for these records.rejection_reasonsincludes Greenhouse’s default reasons, not only those added by your organisation.attachmentsreturns aurlfor each file, which Greenhouse expires after seven days. Each request can redirect to a new short-lived file URL. A replicatedurlstops working one week after the sync that wrote it. Resync the stream to get current links.application_stagescontains one row for each application and stage entered. It therefore holds several times more rows thanapplications. Enable it only for funnel or time-in-stage analysis.candidate_educationsreferences schools, degrees and disciplines by custom field option ID:school_name_custom_field_option_id,degree_custom_field_option_id, anddiscipline_custom_field_option_id. Joinschools,degrees, anddisciplinesto resolve their names.interviewerscontains one row for each panel member and interview. An interview appears once for every attendee.user_idis empty for external attendees who don’t match a Greenhouse user. Their address appears inemailinstead.scorecard_candidate_attributescontains one row for each rated attribute and scorecard. It therefore holds many more rows thanscorecards.scorecard_question_answerscontains one row for each question and scorecard, including unanswered questions. It therefore holds many more rows thanscorecards. The question’sanswer_typedetermines which field contains the answer. Useanswerfortext,boolean_valueforyes_no, andscorecard_question_answer_optionsforsingle_selectandmulti_select.valuestores every answer as text. This can be the text itself,"true"/"false", an option ID or a JSON array of option IDs such as"[26384455003, 26384458003]". For an unanswered question,answeris empty andvalueisnull. Formulti_select, it is"[]".interview_kits.exercisescontains the live Interview Prep content and can include HTML. Thesummaryandinstructionsfields onjob_interviewsare snapshots from when the slot was first added. They don’t reflect later edits. Useexercisesfor the current text.user_emailslists every address for a user, whileusers.primary_emailcontains one. Integration service users have no addresses, so they have no rows here.approvers.send_auto_reminder_atis a date (YYYY-MM-DD), not a timestamp. This differs from every other Harvest v3 field using the_attimestamp suffix.- The 18 streams that became incremental in 1.1.0 now honour Start date. Previously, they always read full history. See Streams that became incremental in 1.1.0.
Deletions
Greenhouse Harvest v3 doesn’t report permanently deleted records. The connector replicates soft deletions using a flag on each record:
activeonjob_posts. Harvest v3 excludes deleted posts by default, so the connector requests active and deleted posts. Filter onactivedownstream if you only need live posts.activeoncustom_fields,custom_field_options,degrees,disciplines,schools,demographics_question_sets,demographics_questions,demographics_question_sets_questions,demographics_answer_options,demographics_answers_answer_options,prospect_pools,scorecard_question_options, andjob_candidate_attributes. Harvest v3 returns active and inactive records for these streams.deactivatedonusers. Deactivated users remain in the stream.activeonjob_stages,job_interviews, andscorecard_questions.falsemeans the stage, interview or question was removed from the job’s interview plan or interview kit. Greenhouse retains the record so historical data still resolves.
Permanently deleted Greenhouse records remain in your destination until you resync the stream in Full Refresh - Overwrite.
Configuration error requests re-authentication
This section applies to OAuth. For Client Credentials, see Client Credentials authentication fails.
The connector can’t renew its access token because Greenhouse rejected the refresh token. From version 1.0.1, this appears as a configuration error instead of a system error. Use the Greenhouse error code in the sync log to identify the fix:
invalid_grant: the refresh token expired or became invalid. This happens when the connection hasn’t synced for over 14 days. It also happens when another tool uses the same refresh token. A sync interrupted immediately after token renewal can cause it too. In the last two cases, Greenhouse issued a new refresh token that Airbyte didn’t save. Greenhouse replaces the refresh token during every renewal. Open the source settings, click Authenticate and complete consent again to store a new token. Run consent separately for every Airbyte source. Don’t reuse a refresh token across sources or other tools.invalid_clientorunauthorized_client: Greenhouse rejected Airbyte’s partner application credentials during refresh. The application might also lack permission to use the refresh token grant. Open the source settings, click Authenticate and complete consent again. If the error continues, contact Airbyte support. You don’t have any credentials to correct.
Client Credentials authentication fails
With Client Credentials, the connector requests a new Greenhouse access token using your client ID and secret. There is no refresh token to renew. The shared OAuth message still says the refresh token was rejected and asks you to re-authenticate. Use the Greenhouse error code at the message’s end instead:
invalid_client: Greenhouse doesn’t recognise the client ID and client secret. Check both against the credential in Configure > Dev Center > API Credential Management. Confirm that the credential hasn’t been revoked.invalid_grant: the Site Admin user ID doesn’t match a Greenhouse user. Correct it or leave it blank. A blank value sends requests as the credential’s integration service user.
Stream fails with a 403 configuration error
The authorising user isn’t a Site Admin, or consent didn’t include the stream’s scope. This is expected when streams added in 1.2.0, streams added in 1.4.0, or streams added in 1.5.0 is enabled on a source authorised before that version. You might have enabled it manually. Airbyte might also enable it during upgrades when the connection propagates all field and stream changes. Compare the scopes with your approved scopes. Then open the source settings, click Authenticate and repeat consent as a Site Admin. With Client Credentials, grant the missing scope to the Greenhouse credential. Confirm that Site Admin user ID is blank or belongs to a Site Admin.
Fewer records after upgrading to 1.1.0
Version 1.1.0 made 18 streams incremental on updated_at. Every sync mode now applies Start date as a Greenhouse query filter for these streams. With a start date, records last updated before it aren’t returned. This applies to incremental and full refresh syncs. If the connection writes in Full refresh | Overwrite, the first post-upgrade sync also deletes older destination rows. Your warehouse therefore shows the gap even though the sync succeeded. Earlier versions didn’t filter these streams by date and ignored Start date entirely. Streams that became incremental in 1.1.0 lists the affected streams.
To read full history again, clear Start date in the source settings. Then refresh the affected streams to write the older records again.
Changelog
Expand to review
| Version | Date | Pull Request | Subject |
|---|---|---|---|
| 1.5.3 | 2026-10-13 | 88668 | Update dependencies |
| 1.5.2 | 2026-10-05 | 87672 | Promote the connector to certified |
| 1.5.1 | 2026-10-05 | 87670 | Update to source-declarative-manifest 7.33.0 and name the fix for each authentication method in the 403 error message |
| 1.5.0 | 2026-10-02 | 87442 | Add the scorecard_question_answers, scorecard_question_options, scorecard_question_answer_options, job_candidate_attributes, and job_post_locations streams, which need new scopes - see Streams added in 1.5.0 |
| 1.4.0 | 2026-09-28 | 86478 | Add the candidate_attribute_types and job_notes streams, which need new scopes - see Streams added in 1.4.0 |
| 1.3.0 | 2026-09-24 | 85178 | Add client-credentials authentication for Greenhouse custom integrations and self-managed deployments. |
| 1.2.0 | 2026-09-21 | 86477 | Add 20 Harvest v3 detail streams, which need new scopes - see Streams added in 1.2.0 |
| 1.1.0 | 2026-09-17 | 85841 | Sync 18 previously full-refresh streams incrementally, so Start date now applies to them in every sync mode, and suggest 10 streams for new connections - see Streams that became incremental in 1.1.0 |
| 1.0.3 | 2026-09-15 | 85507 | Update dependencies |
| 1.0.2 | 2026-09-02 | 85306 | Clarify in the spec that OAuth credentials come from Airbyte’s Greenhouse partner application and must not be requested from Greenhouse |
| 1.0.1 | 2026-09-02 | 85300 | Surface expired or rotated refresh tokens (invalid_grant) as a re-authenticate config error instead of a system error |
| 1.0.0 | 2026-08-28 | 84846 | Breaking migration from Harvest v1 to Harvest v3 with OAuth. See the migration guide. |
| 0.8.1 | 2026-08-18 | 84641 | Update dependencies |
| 0.8.0 | 2026-08-12 | 83811 | Send pagination page-size parameters only on first-page requests and use fully-qualified per-stream URLs in preparation for the Harvest v3 migration. |
| 0.7.33 | 2026-08-11 | 83956 | Update dependencies |
| 0.7.32 | 2026-07-28 | 83194 | Update to CDK 7.23.8 (fixes AirbyteCustomCodeNotPermittedError for bundled custom components) and remove the temporary Cloud version override |
| 0.7.31 | 2026-07-28 | 1082 | Roll Cloud back to 0.7.29 — 0.7.30 is built on SDM 7.23.7, which breaks bundled custom components |
| 0.7.30 | 2026-07-28 | 82944 | Update dependencies |
| 0.7.29 | 2026-07-21 | 82444 | Update dependencies |
| 0.7.28 | 2026-07-14 | 81887 | Update dependencies |
| 0.7.27 | 2026-06-30 | 81129 | Update dependencies |
| 0.7.26 | 2026-06-23 | 80487 | Update dependencies |
| 0.7.25 | 2026-06-16 | 79888 | Update dependencies |
| 0.7.24 | 2026-06-09 | 79354 | Update dependencies |
| 0.7.23 | 2026-06-02 | 78766 | Update dependencies |
| 0.7.22 | 2026-05-15 | 78119 | Set the default concurrency to 2 and expose the number of concurrent threads as a user-configurable option. |
| 0.7.22-rc.3 | 2026-05-12 | 78052 | Reduce default_concurrency to 3 for concurrency tuning after rate-limit failures at higher settings. |
| 0.7.22-rc.2 | 2026-05-08 | 78006 | Concurrency tuning iteration: bump default_concurrency to 5 |
| 0.7.22-rc.1 | 2026-05-06 | 77826 | Start concurrency tuning at default_concurrency=4 (Path A) and enable progressive rollout |
| 0.7.21 | 2026-04-28 | 77287 | Update dependencies |
| 0.7.20 | 2026-04-21 | 76637 | Update dependencies |
| 0.7.19 | 2026-03-31 | 75729 | Update dependencies |
| 0.7.18 | 2026-03-17 | 74919 | Update dependencies |
| 0.7.17 | 2026-03-10 | 74688 | Update dependencies |
| 0.7.16 | 2026-03-03 | 74176 | Update dependencies |
| 0.7.15 | 2026-02-10 | 73107 | Update dependencies |
| 0.7.14 | 2026-02-03 | 72661 | Update dependencies |
| 0.7.13 | 2026-01-20 | 71894 | Update dependencies |
| 0.7.12 | 2026-01-14 | 71700 | Update dependencies |
| 0.7.11 | 2025-12-18 | 70503 | Update dependencies |
| 0.7.10 | 2025-11-25 | 70056 | Update dependencies |
| 0.7.9 | 2025-11-18 | 69420 | Update dependencies |
| 0.7.8 | 2025-10-29 | 68823 | Update dependencies |
| 0.7.7 | 2025-10-21 | 68226 | Update dependencies |
| 0.7.6 | 2025-10-14 | 67896 | Update dependencies |
| 0.7.5 | 2025-10-07 | 67399 | Update dependencies |
| 0.7.4 | 2025-09-30 | 66408 | Update dependencies |
| 0.7.3 | 2025-09-09 | 65896 | Update dependencies |
| 0.7.2 | 2025-08-28 | 64973 | Update dependencies |
| 0.7.1 | 2025-08-26 | 65510 | Fix custom migrations to reference DeclarativeStream Pydantic model instead of runtime component |
| 0.7.0 | 2025-07-07 | 62830 | Promoting release candidate 0.7.0-rc.1 to a main version. |
| 0.7.0-rc.1 | 2025-06-29 | 47283 | Migrate to Manifest-only |
| 0.6.1 | 2025-03-22 | 53800 | Update dependencies |
| 0.6.0 | 2025-03-14 | 55774 | Promoting release candidate 0.6.0-rc.1 to a main version. |
| 0.6.0-rc.1 | 2025-03-14 | 54702 | Update to latest airbyte-cdk, remove custom cursors. |
| 0.5.32 | 2025-02-01 | 52724 | Update dependencies |
| 0.5.31 | 2025-01-25 | 51842 | Update dependencies |
| 0.5.30 | 2025-01-11 | 51214 | Update dependencies |
| 0.5.29 | 2024-12-28 | 50632 | Update dependencies |
| 0.5.28 | 2024-12-21 | 50109 | Update dependencies |
| 0.5.27 | 2024-12-14 | 49248 | Starting with this version, the Docker image is now rootless. Please note that this and future versions will not be compatible with Airbyte versions earlier than 0.64 |
| 0.5.26 | 2024-12-12 | 48996 | Update dependencies |
| 0.5.25 | 2024-10-29 | 47110 | Update dependencies |
| 0.5.24 | 2024-10-23 | 47306 | Add ‘job_post_id’ to applications stream scehma |
| 0.5.23 | 2024-10-12 | 46828 | Update dependencies |
| 0.5.22 | 2024-10-05 | 46506 | Update dependencies |
| 0.5.21 | 2024-09-28 | 46159 | Update dependencies |
| 0.5.20 | 2024-09-21 | 45834 | Update dependencies |
| 0.5.19 | 2024-09-17 | 45625 | Change check stream |
| 0.5.18 | 2024-09-14 | 45476 | Update dependencies |
| 0.5.17 | 2024-09-07 | 45229 | Update dependencies |
| 0.5.16 | 2024-08-31 | 44755 | Update dependencies |
| 0.5.15 | 2024-08-17 | 44246 | Update dependencies |
| 0.5.14 | 2024-08-10 | 43595 | Update dependencies |
| 0.5.13 | 2024-08-03 | 43160 | Update dependencies |
| 0.5.12 | 2024-07-27 | 42816 | Update dependencies |
| 0.5.11 | 2024-07-20 | 42240 | Update dependencies |
| 0.5.10 | 2024-07-13 | 41787 | Update dependencies |
| 0.5.9 | 2024-07-10 | 41215 | Update dependencies |
| 0.5.8 | 2024-07-10 | 39601 | Move spec to manifest, fix readme |
| 0.5.7 | 2024-07-06 | 40882 | Update dependencies |
| 0.5.6 | 2024-06-25 | 40451 | Update dependencies |
| 0.5.5 | 2024-06-22 | 39968 | Update dependencies |
| 0.5.4 | 2024-06-06 | 39247 | [autopull] Upgrade base image to v1.2.2 |
| 0.5.3 | 2024-04-19 | 36640 | Updating to 0.80.0 CDK |
| 0.5.2 | 2024-04-12 | 36640 | schema descriptions |
| 0.5.1 | 2024-03-12 | 35988 | Unpin CDK version |
| 0.5.0 | 2024-02-20 | 35465 | Per-error reporting and continue sync on stream failures |
| 0.4.5 | 2024-02-09 | 35077 | Manage dependencies with Poetry. |
| 0.4.4 | 2023-11-29 | 32397 | Increase test coverage and migrate to base image |
| 0.4.3 | 2023-09-20 | 30648 | Update candidates.json |
| 0.4.2 | 2023-08-02 | 28969 | Update CDK version |
| 0.4.1 | 2023-06-28 | 27773 | Update following state breaking changes |
| 0.4.0 | 2023-04-26 | 25332 | Add new streams: ActivityFeed, Approvals, Disciplines, Eeoc, EmailTemplates, Offices, ProspectPools, Schools, Tags, UserPermissions, UserRoles |
| 0.3.1 | 2023-03-06 | 23231 | Publish using low-code CDK Beta version |
| 0.3.0 | 2022-10-19 | 18154 | Extend Users stream schema |
| 0.2.11 | 2022-09-27 | 17239 | Always install the latest version of Airbyte CDK |
| 0.2.10 | 2022-09-05 | 16338 | Implement incremental syncs & fix SATs |
| 0.2.9 | 2022-08-22 | 15800 | Bugfix to allow reading sentry.yaml and schemas at runtime |
| 0.2.8 | 2022-08-10 | 15344 | Migrate connector to config-based framework |
| 0.2.7 | 2022-04-15 | 11941 | Correct Schema data type for Applications, Candidates, Scorecards and Users |
| 0.2.6 | 2021-11-08 | 7607 | Implement demographics streams support. Update SAT for demographics streams |
| 0.2.5 | 2021-09-22 | 6377 | Refactor the connector to use CDK. Implement additional stream support |
| 0.2.4 | 2021-09-15 | 6238 | Add identification of accessible streams for API keys with limited permissions |