v2.1.1

What's Changed

• chore: point stable marketplace to v1.18.1 by @younnieCutler in #55
• feat(routing): Routing Autoresearch benchmark and experiment runner by @younnieCutler in #56
• feat(routing): Routing Autoresearch agent program and subject capsule by @younnieCutler in #57
• feat(routing): routing-eval-v2 — rebalanced, thickened, version-scoped by @younnieCutler in #58
• feat(career-agent): first autonomous routing research run by @younnieCutler in #59
• feat(career): separate career readiness from job-search intent by @younnieCutler in #60
• feat(career): projects as work-event context, JD reuse, and occasional maintenance by @younnieCutler in #61
• docs: correct the 1.24.0 entry and declare each skill's license by @younnieCutler in #62
• feat: Evidence-First Career OS — 棚卸し to rendered 職務経歴書 (v2.0.0) by @younnieCutler in #63
• feat: rename to japan-career-agent, and install without a plugin host (v2.1.0) by @younnieCutler in #64
• docs: correct the release-channel section, and check it from now on (v2.1.1) by @younnieCutler in #67

Full Changelog: v1.18.1...v2.1.1

v1.18.1

What's Changed

• chore: point stable marketplace to v1.17.2 by @younnieCutler in #51
• fix(policy): skip .worktrees when scanning for forbidden output claims by @younnieCutler in #54
• feat(career-agent): progressive onboarding for new vaults by @younnieCutler in #52
• feat(career-agent): route chuto topics to specific tenshoku references by @younnieCutler in #53

Full Changelog: v1.17.2...v1.18.1

v1.17.2

What's Changed

• feat: harden release and approval durability by @younnieCutler in #50

Full Changelog: v1.17.1...v1.17.2

v1.17.1

What's Changed

• feat: block personal career documents from Git tracking (PRD phase 1) by @younnieCutler in #29
• feat: private career-document store (PRD phase 2) by @younnieCutler in #30
• feat: personal fact timeline and current projection (PRD phase 3) by @younnieCutler in #31
• feat: personal context selection and the downstream read path (PRD phase 4) by @younnieCutler in #32
• feat: promote an imported document to a canonical fact (PRD phase 5) by @younnieCutler in #33
• feat: LLM-as-judge pilot for job-seeker-agent (advisory, 4 synthetic fixtures) by @younnieCutler in #34
• fix: score commentary language and artifact language separately in the judge rubric by @younnieCutler in #35
• fix: repair four rubric defects, then withdraw the pilot verdict — subject output is non-deterministic by @younnieCutler in #36
• fix: make Conflict a Decision Status value only, not a requirement state (1.12.1) by @younnieCutler in #37
• feat: add Career Agent P0 UX contract by @younnieCutler in #38
• feat: add P1 progressive disclosure and executable workflows by @younnieCutler in #39
• feat: add P1 guided thin frontend by @younnieCutler in #40
• test: verify the judge pilot's subject before its treatment, and stop where the control moves by @younnieCutler in #41
• test: add UX regression eval and calibration by @younnieCutler in #42
• fix: state the Missing/Unknown boundary in SKILL.md, where every run reads it (1.16.1) by @younnieCutler in #43
• test: re-verify the subject under 1.16.1 and isolate what still moves by @younnieCutler in #44
• test: pre-register a stable control and a single-invariant injection for judge round 4 by @younnieCutler in #45
• test: record LLM judge pilot round 4 by @younnieCutler in #46
• docs: close LLM judge pilot without adoption by @younnieCutler in #47
• test: run six autoresearch-style LLM judge v2 experiments by @younnieCutler in #48
• feat: localize Career Agent UX by message language by @younnieCutler in #49

Full Changelog: v1.7.1...v1.17.1

v1.7.1

What's Changed

• refactor: extract career agent persistence and vault by @younnieCutler in #18
• refactor: extract career agent routing by @younnieCutler in #19
• refactor: extract career agent proposals and lifecycle by @younnieCutler in #20
• refactor: extract career agent workspace projection by @younnieCutler in #21
• refactor: complete career agent runtime boundary by @younnieCutler in #22
• v1.7 PR7: deterministic behavior eval runner by @younnieCutler in #23
• v1.7 PR8: execute critical behavior scenarios by @younnieCutler in #24
• v1.7 PR9: pin dependencies and add SBOM by @younnieCutler in #25
• v1.7 PR10: harden release integrity by @younnieCutler in #26
• v1.7 PR11: endpoint health canary and v1.7.0 by @younnieCutler in #27
• fix: deadline calendar validation, numeric evidence boundary, MHLW allowlist, Actions SHA pin by @younnieCutler in #28

Full Changelog: v1.6.12...v1.7.1

v1.6.12

What's Changed

• refactor: extract career agent model contracts by @younnieCutler in #17

Full Changelog: v1.6.11...v1.6.12

v1.6.11

What's Changed

• feat: make mock interview probing adaptive by @younnieCutler in #16

Full Changelog: v1.6.9...v1.6.11

v1.6.9

What's Changed

• feat: harden reproducible E2E artifact packaging by @younnieCutler in #15

Full Changelog: v1.6.7...v1.6.9

v1.6.7

What's Changed

• fix: harden evidence and lifecycle E2E (v1.6.7) by @younnieCutler in #14

Full Changelog: v1.6.6...v1.6.7

v1.6.6

What's Changed

• refactor: split career-agent entrypoint boundaries by @younnieCutler in #13

Full Changelog: v1.6.5...v1.6.6