AI Job Search changelog: what's new each month
Every stable AI Job Search release summarised by month: the highlights, new features, improvements, fixes and anything you need to act on. 3 months covered; the current month updates daily.
November 2026 · so far
3 releases: 1.7.4 → 2.0.0
November brought bug fixes across job search, ranking, and data syncing, plus a major version bump to 2.0.0 with new security and employment type features.
Highlights
- Employment type is now recorded and displayed for job postings.
- Security guards can now be customized locally via a new configuration file.
- Better handling of combined columns in salary workbooks.
- Improved error reporting replacing tracebacks with user-friendly messages.
- Fixed region classification so the US is recognized as a country, not a region.
New
- Fork-local security guard negations via .github/security-guards-local.txt.
- Recording and surfacing employment type for job postings.
Improved
- Freehire search now honors Retry-After headers on rate limit responses.
- Better documentation of OR and title: operators in search queries.
- Notion sync re-keys rows when the Key column has drifted but URL still matches.
- Installer now finds bun at standard locations even when it's missing from PATH.
- Fetch-page passes --compressed flag only when curl supports it.
- Scrape output forced to UTF-8 encoding.
- Job title deduplication now keeps C++, C#, and C distinct.
- Setup now clears inferred writing-style patterns on reset.
Fixed
- Freehire search rejects detail URLs on other hosts instead of querying the API.
- Job key audit reports now show non-string URLs without tracebacks.
- Rank apply no longer crashes on malformed scoring results.
- Rank no longer crashes on non-object entries in seen_jobs.json.
- Latin letters without NFKD decomposition now fold correctly with --audit --migrate.
- Zlib-wrapped HTTP deflate responses are now decoded properly.
October 2026
1 release: 1.7.3
October focused on security hardening and stability improvements for the AI Job Search agent. Several fixes addressed edge cases in data handling and posting retrieval.
Highlights
- Security improvements to skill and command permissions with stricter allowlist checking.
- Default deny rules now block unsafe upload methods and protect sensitive credential files.
- Posting-fetcher, posting-scorer and application-reviewer agents restricted to read-only access.
Improved
- Page fetching now checks robots.txt before making requests.
- Better handling of Unicode whitespace in skill grouping.
- Improved support for non-finite values in text cells.
- More robust handling of edge cases like whitespace-only company names.
Fixed
- 403 retry logic in page fetching now uses documented behaviour.
- Settings.json permission rules now apply consistently across all permission modes.
- Chart-only tabs now render correctly.
September 2026
2 releases: 1.7.1 → 1.7.2
September brought stability improvements and better handling of job posting details. The team focused on data persistence, URL parsing robustness, and fixing edge cases in salary extraction.
Highlights
- Job posting publication dates are now captured and used by the ranking system.
- LinkedIn job URLs with trailing slashes, regional subdomains, and tracking parameters are properly handled.
- Salary converter now correctly handles Count/Index column pairs without false data splits.
- Interview prep documentation path clarification eliminates misleading ignore rules.
New
- CHANGELOG structure validation on every pull request to catch formatting errors early.
Improved
- LinkedIn search detail parsing now accepts a wider variety of LinkedIn job URL formats.
- Salary converter has real Excel workbook integration tests for more reliable extraction.
- Better error messaging in the layout verification tool to identify both causes of PDF extraction failures.
Fixed
- Salary data no longer generates false "too few employees" rows from bare Count/Index column pairs.
- Layout verifier now properly reports when Poppler versions 26.0x before 26.05 fail on PDFs with empty Info fields.
Summaries are written automatically from the official release notes (full changelog ↗); check the original notes before relying on a detail. AI Job Search: pricing, features and alternatives · All changelogs