The newest remote jobs from Remote OK and Jobicy in one deduplicated list, with salary, location, tags and the apply link. Filter by keyword, location, minimum salary and posting date, or get only new jobs on a schedule. $0.001 per job ($1 per 1,000); duplicates and filtered jobs are free.
2026-11-14 r3: competitor price table removed from the README
README only (Mike 2026-11-14): the "How it compares" table of competitor Actors and their listed prices is replaced by a short feature list with no prices and no competitor names. Links to our own Actors are kept. No code, schema, price, default or output change.
New guard test tests/test_1114_no_competitors.py (no third-party Actor links, no known competitor names, no competitor price-table wording).
2026-11-14 r2: copy wording (next build of 1.5)
README: 'official public feeds' reads 'public feeds'; 'one alert across three remote boards' reads 'one list across two remote boards' (also a fix: the Actor reads two boards).
Copy (Lex's rules, Mike 2026-11-14): no 'official', 'monitor(ing)' or 'alert' wording left in README, .actor/actor.json or .actor/input_schema.json. 'Monitoring mode' is now 'only-new mode (onlyNewJobs)' on scheduled runs; the onlyNewJobs field title is 'Only new jobs' and the section caption 'Only-new runs and output'. Text only: no field name, default, prefill, price or output change. New guard test tests/test_1114_copy_words.py.
Files: README.md, .actor/input_schema.json, CHANGELOG.md, tests/test_1114_copy_words.py (new). No code change.
2026-11-13: chain dataset readable under limited permissions (next build of 1.5)
Input schema:datasetId now declares "resourceType": "dataset" and "resourcePermissions": ["READ"] (editor stays textfield, field name and type unchanged). Apify then grants a limited-permissions run read access to the dataset given there. Fixes the chain run that failed with ForbiddenError ... Insufficient permissions on GET /v2/datasets/<id> (2026-11-13 cloud smoke of career-site-jobs-feed 0.3.1). The field description now says so and suggests {{resource.defaultDatasetId}} for integrations.
Code (my_actor/jobschema.py): reads the dataset through the declared datasetId field as before; the integration payload resource.defaultDatasetId fallback is kept. When the dataset cannot be read, the error message now says which of the two it came from, and for the payload it says to put {{resource.defaultDatasetId}} in datasetId. Same error type, same point in the run, nothing fetched or charged.
README: "Chain with other jobs Actors": datasetId row mentions {{resource.defaultDatasetId}}; one new bullet on read-only access under limited permissions.
No change to output, charging, prices, defaults, prefill or runs without datasetId.
2026-11-13: README rebased onto live 1.4.1
README only. Restored the live Related Actors wording: ATS Jobs Scraper covers 9 systems including Teamtailor, Polymer and HireHive. The "Chain with other jobs Actors" section is unchanged. No code, schema, input or price change.
2026-11-11: chain with other jobs Actors (datasetId), shared job format (version 1.5)
New optional inputs datasetId, datasetMode, datasetMatchBy, suiteOutput (input-schema section "Chain with other jobs Actors (advanced)", no schema defaults, no prefill). Without them the run is unchanged: same items, same field order, same values, same charging, same RUN_SUMMARY keys (checked against golden output recorded from version 1.4 in tests/fixtures/suite_golden_current.json).
datasetId: job items of that dataset are read (in pages, at most 100,000 items) before any board; a job already there (by uid, or also by crossSourceKey with datasetMatchBy "uidOrCrossKey") is dropped before the push and never charged. datasetMode "merge" (default) copies the earlier job items first, free (pushed without an event name); "exclude" copies nothing. Status items and expired/unresolved records are not copied. Falls back to the Apify integration payload resource.defaultDatasetId when datasetId is empty. maxTotalJobs counts only new jobs.
Unreadable datasetId, unknown datasetMode or datasetMatchBy: the run ends with one free status item (status "error") and no jobs, as for other invalid input in this Actor.
suiteOutput: true: each job item is rewritten into the shared field order (my_actor/jobschema.py, SUITE_FIELDS), with missing shared fields as null or []; values unchanged; jobs copied from datasetId are kept as they were. Off by default.
RUN_SUMMARY gets previousDataset (only when datasetId is used) and suiteOutput (only when true).
New shared file my_actor/jobschema.py (JSON Schema jobs-suite/1, byte-identical in all four jobs Actors). Tests: tests/test_1111_suite_contract.py, tests/test_1111_suite_chain.py, tests/suite_mock.py, fixtures jobs-suite-item.schema.json, suite_golden_current.json, suite_shared.sha256.json.
README: section "Chain with other jobs Actors". No change to prices, requirements.txt (apify >= 4.1, < 4.2), the Dockerfile, exampleRunInput or the prefill.
2026-11-11: dedupe off keeps every copy, jobicyTag wording from Jobicy's docs, Jobicy 100 per request (version 1.4)
Bug fix: dedupe: false now keeps every copy. Before (since at least 1.1.1, still in 1.3), the merge step stored jobs by company + title even with dedupe off, so a later job with the same normalized company and title replaced the earlier one: only one copy (the older one, after the newest-first sort) was pushed, and the newer copy was silently dropped and never charged. Now, with dedupe off, every job read is kept and goes through the turns, filters, cap and charging like any other job (each copy is one job event). duplicatesMerged stays 0 and alsoListedOn stays empty with dedupe off. With dedupe on (the default), nothing changes.
onlyNewJobs with dedupe off: the company + title key is now checked against what earlier runs saved, not against copies pushed earlier in the same run, so all cross-posted copies found in one run are pushed (before the fix, only one copy reached that check anyway). The uid check is unchanged (a job id already pushed in this run is still skipped). What is saved for the next run is unchanged (uid and company + title key of each pushed job), so the next run skips every copy. With dedupe on, unchanged.
jobicyTag wording: README row and input-schema description now say what Jobicy's own API docs say about tag (keyword search; the API page says job title, the OpenAPI spec 2.2.19 says available job content, 3-50 characters ASCII). Code unchanged: the value is still sent unchanged as tag, not length-checked, not filtered on.
Behaviour change: Jobicy is asked for 100 jobs per request, not 200 (count=100 in my_actor/feeds.py). 100 is the stricter documented limit: Jobicy's API page (jobs-rss-feed) says count accepts 1-100 and defaults to 100; its OpenAPI spec 2.2.19 says 1-200. Paging is unchanged: the adapter follows nextCursor with the same filters while hasMore is true, for at most 50 requests, so caps above 100 are still reached through cursor pages when Jobicy has enough matching jobs. Effects: the most Jobicy jobs one run can read drops from 10,000 (50 x 200) to 5,000 (50 x 100); reading more than 100 Jobicy jobs now takes twice as many requests (e.g. 300 jobs: 3 requests instead of 2). Runs that need at most 100 Jobicy jobs are unchanged (the default 8-job run reads 16 at most and still makes one Jobicy request). Remote OK unchanged.
Tests: tests/test_1111_interleave.py: the strict xfail test_dedupe_off_keeps_every_copy is now a passing test; new test_dedupe_off_with_only_new_keeps_every_copy_in_one_run and test_dedupe_off_many_copies_and_cap. tests/test_1111_jobicy_tag.py expects count 100. New tests/test_1111b_jobicy_paging.py (fake Jobicy: 100 per page, cursor pages keep the filters, 50-page / 5,000-job limit, stop on hasMore false, a 150 cap reached through cursor pages, the default run still one request).
.actor/actor.json version is 1.4. No change to prices, requirements.txt (apify >= 4.1, < 4.2), the Dockerfile, exampleRunInput or the prefill.
2026-11-11: boards take turns before the job cap, jobicyTag documented (version 1.3)
Behaviour change: boards take turns (round-robin) before maxTotalJobs is applied. Before, all jobs were put in one newest-first list and the first maxTotalJobs were pushed, so one board could fill a small cap. Ranger's 1.2.1 default run (uHnOiqRbvpgCcQcSF) pushed 8 Jobicy jobs and 0 Remote OK jobs although Remote OK was read (16 jobs, status ok), because every Jobicy job was newer. Now the jobs are handed out one board at a time in sources order (Remote OK first when sources is empty or in the default order), newest first within each board, and the cap stops the push as before. With both boards readable and no cross-posted duplicates, the 8-job default is 4 Remote OK + 4 Jobicy, alternating.
When a board has run out of matching jobs (none fetched, fewer than its share, filtered out, or Jobicy skipped by its hourly fair-use rule), the other board fills the rest of the cap.
Filters (keywords, excludeKeywords, locations, requireSalary, minSalaryUsd, postedWithinDays, and the salary read from description text) are applied before a job takes a turn, so a job that does not match never uses up a board's slot. A job skipped by onlyNewJobs (seen in an earlier run) is skipped where it falls in the turn order, as before.
Dataset order changes: items now alternate by board instead of being strictly newest first across boards. Each board's own items are still newest first.
The salary read from the description text (salary_from_text) now runs lazily, as each job is reached in the turn order, instead of in newest-first order; the values are the same.
Unchanged: the merge of cross-posted duplicates (same normalized company and title) still runs first on the newest-first list, so the newest copy is kept and the others go to alsoListedOn; duplicatesMerged; uid, dedupeKey, crossSourceKey; the new-only state keys; the per-board early stop at 2x the cap; charging (one job event per pushed job, default run still at most 8 x $0.001 = $0.008); log and status texts, including Default mode: no sources or filters given, returning up to 8 of the newest jobs from ...; the Remote OK and Jobicy adapters.
jobicyTag documented (README input table, and a new jobicyTag field in .actor/input_schema.json: type string, text field, no default, no prefill). Verified in my_actor/feeds.py: a non-empty value is passed unchanged to Jobicy's API as tag (with geo / industry when those are set); an empty value is not sent; Remote OK ignores it; this Actor does not filter on it. The code path is unchanged; the input already worked from the API. Side effect of the schema entry: the field now shows in the Console form, and the platform's input validation now expects a string, so a non-string jobicyTag sent through the API would be rejected by the platform before the run starts (before, a list was passed through to Jobicy).
README: default-mode paragraph and the maxTotalJobs row describe the turn order; new jobicyTag row.
Tests: tests/test_1111_interleave.py (new: unit tests of interleave_sources; whole runs on fake boards: both boards present at cap 8 for {} and the prefill, sources order decides who goes first, one board empty so the other fills, a short board runs out then the other fills, dedupe across the interleave, filters applied before the turns; plus one strict xfail that records a pre-existing dedupe: false bug, see below), tests/test_1111_jobicy_tag.py (new). tests/test_1110_nowwr.py: the comparison with the 1.1.1 _base/ tree is now skipped with a clear reason when _base/ is not a sibling folder (a bare Actor tarball); when _base/ is present it runs unchanged. tests/test_1103_tester.py: the unused workday-jobs-scraper sample input URL is now https://utaustin.wd1.myworkdayjobs.com/UTstaff (test-only, no behaviour change).
Known, not changed in this version: with dedupe: false, cross-posted jobs with the same normalized company and title are still collapsed by the merge (only the older copy is pushed), although the README says turning dedupe off keeps every copy. Present since before 1.2.1; recorded by the strict xfail test.
.actor/actor.json version is 1.3. No change to prices, requirements.txt (apify >= 4.1, < 4.2), the Dockerfile, exampleRunInput or the prefill.
2026-11-10: We Work Remotely removed (version 1.2)
We Work Remotely is no longer fetched. Its API terms forbid building a job search service. The board was removed from my_actor/feeds.py (ADAPTERS / SOURCES are now remoteok and jobicy only). The Remote OK and Jobicy adapters are unchanged (function bodies byte-identical to 1.1.1).
If a saved input or schedule still lists weworkremotely in sources, the run does not fail. That name is ignored, one free status item explains the terms reason, and RUN_SUMMARY.sources.weworkremotely is status: "removed". Other requested sources still run. weworkremotely stays in the input-schema enum (titled "We Work Remotely (no longer read)") so old Console inputs are not rejected. wwrCategories is kept and documented as unused.
Default / prefill: empty input and the prefilled form used to read all three boards and stop at 8 jobs ($0.008). Ranger's 1.1.1 smoke (iLqLmBMcZGsVaFwlu) requested Remote OK + WWR and pushed 8 WWR jobs (Remote OK had been fetched, but the 8-job cap was filled from WWR). With WWR gone, the default, the schema prefill and exampleRunInput are Remote OK and Jobicy, still maxTotalJobs 8, still at most $0.008, under the $0.01 public-default cap. The dataset is not left empty: if only weworkremotely is named, the free status item is written; if Jobicy is deferred, Remote OK can still fill the cap.
.actor/actor.json version is 1.2. Title and description no longer list WWR as a current source. No price change ($0.001 per job).
2026-11-10: contact details removed from descriptions, dedupeKey and crossSourceKey (version 1.1)
New input redactContacts (default true): personal e-mail addresses (role mailboxes such as jobs@ or careers@ stay), phone numbers with 8 or more digits (dates stay) and 'Recruiter: name' / 'Hiring manager: ...' lines are replaced with [email removed], [phone removed] and [contact removed] in descriptionText and descriptionHtml (HTML: also mailto: and tel: links). The rules are ats-jobs-scraper's (my_actor/contacts.py), including its 2026-11-10 fix for (555) 123-4567. Only matters when includeDescription is on (default off). Redaction runs after the filters, so searchInDescription still matches the original text.
New output fieldsdedupeKey and crossSourceKey (recipe in my_actor/jobkey.py, byte-identical in ats-jobs-scraper and workday-jobs-scraper).
Unchanged: uid (source:id), the in-run merge of cross-posted jobs (alsoListedOn, still on the old company + title key), the keys saved in the new-only state (uid and the old key, so saved state keeps working), charging (one job event per pushed job), the default run.
README: API example uses /v2/actors/ from one APIFY_API variable; rows for the new fields and redactContacts.
Changed files: .actor/actor.json (version 1.1), .actor/dataset_schema.json, .actor/input_schema.json, CHANGELOG.md, README.md, my_actor/contacts.py (new), my_actor/jobkey.py (new), my_actor/main.py, tests/fixtures/jobkey_vectors.json (new), tests/test_1110_jobkey.py (new), tests/test_1110_redact_keys.py (new). No change to requirements.txt (apify >= 4.1, < 4.2), the Dockerfile or prices.
Unreleased: inputs that crashed the run now get the invalid-input status row (1104d)
postedWithinDays far above the schema maximum (365), e.g. 1e30 or Infinity, used to crash the run with a traceback (exit 91, FAILED, empty dataset). Now the run ends SUCCEEDED (exit 0) with the existing free invalid-input status row (status: "error", message Invalid input, nothing was fetched or charged: postedWithinDays must be a whole number of days up to 365, not ...). Nothing is fetched, no RUN_SUMMARY is written, no result event is charged. Values that worked before (including 366 and above, as long as they did not overflow) are unchanged.
keywords, excludeKeywords and locations (not sources, which is unchanged): an object or list as a list item (e.g. keywords: [{"a": 1}]), or an object instead of a list, now gets the same status row. Before, it was turned into text ("{'a': 1}") or its keys were used, silently. Numbers in a list are still read as text.
An input that is not a JSON object (a list, string or number) now gets the same status row. Before, it was silently run as {} (the default demo). No input at all is still run as {}.
1104d changed these files: CHANGELOG.md, my_actor/main.py, tests/test_1104d_invalid_input.py. No change to prices, events, .actor/ files, README or requirements (apify >= 4.1, < 4.2 was already pinned).
Unreleased: changelog file lists and clearer time-limit log (GAP13)
Log text after the hard time limit now says exactly what happens. Jobs from sources that finished are kept (a source stopped earlier by the soft time budget keeps the jobs it had read); a source still being read when the limit cut it off is dropped, and none of its jobs are pushed or charged. The old line (kept the jobs read so far) suggested that the jobs read from a cut source were kept. Behaviour is unchanged. The warning names the cut sources, and their RUN_SUMMARY error is now stopped: time budget reached while the source was being read; its jobs were dropped.
Tests added for the cut-off log text, the RUN_SUMMARY error, and a run where one source finishes and one is cut mid-read: only the finished source's rows are pushed and charged.
GAP13 changed these files: CHANGELOG.md, my_actor/main.py, tests/test_gap13_cutoff.py. No change to .actor/input_schema.json or .actor/actor.json.
GAP12 changed these files: .actor/actor.json, CHANGELOG.md, README.md, my_actor/common.py, my_actor/main.py, my_actor/runguard.py, tests/test_gap12_body_cap.py, tests/test_gap12_example_input.py, tests/test_gap12_runguard.py. In .actor/actor.json GAP12 only added exampleRunInput (shown in the GAP12 entry below); it did not change the description, the version or any other field. GAP12 made no change to .actor/input_schema.json (defaults, prefill and field texts are as before).
Unreleased: fail soft and a time budget for the Store test (GAP12)
Fail soft: when no job was read (in the demo, or when every source failed or was skipped, for example by the Jobicy hourly fair-use skip), the run pushes one free status item (no charged event) with the reason, logs a warning, and ends SUCCEEDED.
The state store is opened only when onlyNewJobs is on or Jobicy is a source, and store errors are warnings. A response body is streamed and dropped above 64 MB (16 MB in the demo).
The Jobicy fair-use hint in the status message is now shown only when Jobicy really was skipped for fair use.
Time budget: the demo run ({}, the prefill or the exampleRunInput) starts no new request after about 40 s, uses a 10 s timeout per request with at most 2 attempts and waits at most 2 s between them (a long Retry-After is capped). Every run, demo or not, also stops starting requests 20 s before the platform timeout (ACTOR_TIMEOUT_AT) and gets cut off at most 10 s after that, so a slow or hung source can no longer make the run time out. When the platform aborts the run, no new request is started.
robots.txt is now fetched with one lock per host instead of one lock for all hosts, so one slow host no longer holds up the others.
New exampleRunInput in actor.json: {"sources": ["remoteok", "weworkremotely", "jobicy"], "maxTotalJobs": 8}. It matches the prefill and runs the same small demo.
Tests added for a source that is down, slow or hung, a partial failure, the time budget, the free status item and an unreadable state store.
Unreleased: small, fast default run (GAP8)
BREAKING (default change):maxTotalJobs now defaults to 8 (was 0, no limit), in the input schema and in code when the field is left out. Set it to 0 for the old full pull. Before this, an empty input pushed about 2,000 jobs and took about 4 minutes. If you used a pre-release build whose default was 50, the default is now 8, so runs that left the field empty return 8 jobs instead of 50.
BREAKING (output default change):includeDescription now defaults to false (was true), so descriptionText and descriptionHtml are no longer in the output unless you turn it on.
With a maxTotalJobs cap and no filters or monitoring mode, each board now stops after 2x the cap instead of reading every page (Jobicy then needs one request instead of up to 50). The boards list newest first, so the result is the same newest jobs; per-source found counts in RUN_SUMMARY are smaller and show stoppedEarly.
Default mode: when no sources and no filters are given (or all three sources, as in the form prefill) with the default cap, the log says Default mode: ..., RUN_SUMMARY has defaultMode: true, and the status message starts with Default mode (newest 8 jobs, no filters): .
Lint only: removed unused imports in common.py and feeds.py (pyflakes clean). No behaviour change.
2026-10-29: one vocabulary for employmentType
Breaking for exact-value filters:employmentType is now always one of the schema.org values FULL_TIME, PART_TIME, CONTRACTOR, TEMPORARY, INTERN, VOLUNTEER, PER_DIEM, OTHER, or null, the same vocabulary as the ATS job scrapers. Before, each board's own spelling was passed through (Full-Time, Full time, Contract).
The board's own text is kept in the new field employmentTypeRaw. When a board lists several types, the more specific one wins (Full-Time, Contract is CONTRACTOR). A value that fits none of the types (such as Permanent) becomes OTHER.
Remote OK's salary fields are meant to be yearly USD, but some employers enter an hourly rate there. A Remote OK range whose top is $500 or less is now labelled HOUR in salaryPeriod; everything else stays YEAR.
2026-10-29: robots.txt on every redirect
Redirects are now followed one hop at a time (up to 5), and robots.txt is checked for every hop. A host whose robots.txt can't be read (a server error after one retry, a network failure, or an HTML page served as robots.txt) is skipped for the run with a warning. A missing robots.txt (any 4xx) still means no rules.