Stats
Actions
Tags
From careerproof-skills
Build and run custom hiring evaluation models — upload example strong/weak CVs, have AI infer scoring rubrics, customize evaluation dimensions and weights, then batch-evaluate candidates against your bespoke criteria. Use this skill whenever someone wants to create a custom scoring model, build a tailored rubric, define their own evaluation criteria, train a hiring model on examples, upload annotated CVs to teach the system what "good" looks like, or evaluate candidates against company-specific standards. Also triggers on "custom eval", "bespoke rubric", "our own scoring criteria", or "teach it what we look for". This is different from atlas-gem (standard 10-factor scoring) — use this when the user wants THEIR OWN criteria, not the default framework.
How this skill is triggered — by the user, by Claude, or both
Slash command
/careerproof-skills:atlas-custom-evalThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Create custom evaluation models by teaching the AI what good and bad candidates look like, then batch-evaluate candidates against your bespoke rubric.
Create custom evaluation models by teaching the AI what good and bad candidates look like, then batch-evaluate candidates against your bespoke rubric.
atlas_list_custom_eval_models to check for existing modelsatlas_create_custom_eval_model with name and optional descriptionmodel_idUpload artifacts that teach the model what strong, weak, and mixed candidates look like:
File artifacts — Call atlas_upload_custom_eval_artifact with:
model_idfile — Base64-encoded PDF or DOCXartifact_type — One of: cv_with_notes, template, free_text, jdlabel — One of: strong, weak, mixedfilename, notesText artifacts — Call atlas_add_custom_eval_text_artifact with:
model_idartifact_type — One of: cv_with_notes, template, free_text, jdtext_content — The text to addlabel — One of: strong, weak, mixedGuidelines:
Verify uploads with atlas_list_custom_eval_artifacts.
Call atlas_infer_custom_eval_rubric with model_id
Call atlas_get_custom_eval_rubric to review the generated rubric
Present the rubric to the user for review
If adjustments needed: Call atlas_set_custom_eval_rubric_overrides with model_id and overrides object to:
If want to start over: Call atlas_clear_custom_eval_rubric_overrides to revert to AI-inferred rubric
Optional: Call atlas_start_custom_eval_inference (5 credits, async) for AI-generated evaluation dimensions
careerproof_task_status every 5-10s until completedYou need a hiring context with uploaded candidates. If none exist, direct user to /atlas-onboard.
Call atlas_start_custom_eval_batch with:
custom_model_id — The model to evaluate againstcandidate_ids — Array of candidate IDscontext_id — The hiring contextdetail_level (brief/standard/deep)Confirm the credit cost: 5 credits per candidate
Poll atlas_get_custom_eval_batch_status with batch_id every 5-10 seconds until completed
Call atlas_get_custom_eval_batch_results with batch_id for results
Present ranked results with:
/atlas-gem for standard competency scoring/atlas-fit for CV-to-JD alignment/atlas-shortlist for batch GEM + FIT analysis| Action | Cost |
|---|---|
| Create/manage model | FREE |
| Upload artifacts | FREE |
| Infer rubric from artifacts | FREE |
| Customize rubric overrides | FREE |
| AI dimension inference | 5 credits |
| Batch evaluation | 5 credits/candidate |
Example: Model setup (FREE) + evaluate 8 candidates = 40 credits
strong/weak/mixed) — this is the #1 factor in rubric qualityatlas_list_custom_eval_models to find existing onesatlas_update_custom_eval_model to update model metadata without rebuildingatlas_delete_custom_eval_model to clean up models you no longer neednpx claudepluginhub careerproof-labs/careerproof-skills --plugin careerproof-skillsPerforms expert HR assessment of candidates against job postings using custom scoring rubrics, domain knowledge, and resume analysis.
Screens job candidates for high agency, grit, resilience, impact, and technical depth. Useful when reviewing resumes, screening applicants, or evaluating candidates.
Scores a resume against a job description for ATS keyword match, formatting compliance, and narrative strength. Fires automatically when a resume and job description appear together, or on explicit request via '/resume-score'.