Skip to content

feat(apply): requirement-coverage rules and CV targeting improvements from output benchmark#184

Merged
MadsLorentzen merged 1 commit into
masterfrom
feat/benchmark-methodology-improvements
Jul 18, 2026
Merged

feat(apply): requirement-coverage rules and CV targeting improvements from output benchmark#184
MadsLorentzen merged 1 commit into
masterfrom
feat/benchmark-methodology-improvements

Conversation

@MadsLorentzen

Copy link
Copy Markdown
Owner

Seven methodology improvements from a private output-quality benchmark: the current framework was re-run end-to-end on three real historical applications and blind-scored against the actually-submitted packages. It won 3-0 - but the judges' regression probes identified concrete things the older outputs did better, and this PR folds those back in:

apply.md - new Requirement Coverage block (Step 2):

  1. Every stated requirement addressed - matched or honestly gapped, never silently omitted (the benchmark's own run omitted a stated Kubernetes requirement; a blind judge called the omission 'hiding rather than acknowledging')
  2. Nice-to-haves engaged by name with honest adjacency framing
  3. Posting's literal term over synonyms, including CV section headings ('MLOps' beats 'ML Deployment' for ATS and skim-readers)
  4. Stated logistics/prerequisites addressed in the letter: clearances, availability, job/reference ID, and language-capabilities mapping for multi-country employers

05-cv-templates.md:
5. Domain-transfer argument leads the profile statement for domain-changers (the strongest card, played first)
6. Literal-keyword heading guidance under Core Competencies
7. New Evidence Links note - \href on every verifiable named artifact

All additions are honesty-compatible by construction: each rule contains its own truthfulness constraint. framework_version 05 → 1.2.0.

Verification: lint, version guard, 119 tests pass; personal-data scan of diff clean.

🤖 Generated with Claude Code

… from output benchmark

Seven improvements sourced from blind regression probes comparing current
outputs against real April-2026 baselines (each an area where the older
outputs scored better):
- every stated requirement addressed - matched or honestly gapped, never
  silently omitted (the benchmark run omitted a stated Kubernetes
  requirement entirely; omission reads as hiding under questioning)
- nice-to-haves engaged by name with honest adjacency framing; posting's
  literal term preferred, including in CV section headings
- stated logistics/prerequisites addressed in the letter (clearances,
  availability, job ID, multi-country language mapping)
- domain-transfer argument leads the CV profile statement for
  domain-changers
- evidence links (href) on every verifiable named artifact
framework_version: 05 -> 1.2.0.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
@MadsLorentzen
MadsLorentzen merged commit 122db05 into master Jul 18, 2026
12 checks passed
@MadsLorentzen
MadsLorentzen deleted the feat/benchmark-methodology-improvements branch July 18, 2026 20:22
ThomasHjartarson pushed a commit to ThomasHjartarson/ai-job-search-no that referenced this pull request Jul 24, 2026
… from output benchmark (MadsLorentzen#184)

Seven improvements sourced from blind regression probes comparing current
outputs against real April-2026 baselines (each an area where the older
outputs scored better):
- every stated requirement addressed - matched or honestly gapped, never
  silently omitted (the benchmark run omitted a stated Kubernetes
  requirement entirely; omission reads as hiding under questioning)
- nice-to-haves engaged by name with honest adjacency framing; posting's
  literal term preferred, including in CV section headings
- stated logistics/prerequisites addressed in the letter (clearances,
  availability, job ID, multi-country language mapping)
- domain-transfer argument leads the CV profile statement for
  domain-changers
- evidence links (href) on every verifiable named artifact
framework_version: 05 -> 1.2.0.

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant