Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Conduct and document professional reference checks for final-stage candidates. Use when verifying candidate backgrounds before extending offers.
.claude/skills/migrateforce-reference-checker/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 77% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 75% | 0% |
This skill streamlines the reference check process by generating structured questionnaires, scheduling calls with references, documenting responses, and synthesizing findings into actionable hiring recommendations.
| Field | Type | Required | Description | |-------|------|----------|-------------| | candidate_name | string | yes | Candidate being evaluated | | job_title | string | yes | Role they're being considered for | | references | array | yes | List of reference contacts | | key_competencies | array | yes | Skills/behaviors to verify | | concerns | array | no | Specific areas to probe | | check_type | enum | no | standard, executive, technical |
| Field | Type | Description | |-------|------|-------------| | reference_summaries | array | Summary from each reference | | competency_scores | object | Rating per competency | | red_flags | array | Concerns identified | | recommendation | enum | strong_hire, hire, no_hire, needs_discussion | | full_report | document | Complete reference check report |
Standard Questions:
Role-Specific Questions:
Outreach Email:
Subject: Reference Request for [Candidate Name]
Dear [Reference Name],
[Candidate] has applied for a [Role] position at [Company] and provided
your name as a professional reference.
We'd appreciate 15 minutes of your time to discuss their qualifications.
Would any of these times work?
[Time slot 1]
[Time slot 2]
[Time slot 3]
If you prefer, you can complete a brief written questionnaire instead: [link]
Thank you for your time.
[Recruiter name]Call Structure (15 min):
Probing Techniques:
json{ "reference_name": "Jane Smith", "relationship": "Direct manager, 2 years", "company": "Previous Corp", "date_checked": "2025-02-01", "responses": { "strengths": ["Technical depth", "Mentorship"], "growth_areas": ["Public speaking", "Delegation"], "specific_example": "Led migration project...", "would_rehire": true, "rehire_enthusiasm": "Absolutely, without hesitation" }, "overall_sentiment": "strongly_positive" }
Competency Matrix: | Competency | Ref 1 | Ref 2 | Ref 3 | Avg | |------------|-------|-------|-------|-----| | Technical Skills | 5 | 5 | 4 | 4.7 | | Communication | 4 | 4 | 5 | 4.3 | | Leadership | 3 | 4 | 4 | 3.7 | | Reliability | 5 | 5 | 5 | 5.0 |
Red Flag Categories:
markdown# Reference Check Report **Candidate:** [Name] **Position:** [Role] **Date:** [Date] **Checked by:** [Recruiter] ## Summary [2-3 sentence executive summary] ## References Contacted ### Reference 1: Jane Smith - **Relationship:** Direct manager (2019-2021) - **Company:** Previous Corp - **Key Feedback:** [Summary] - **Sentiment:** ⭐⭐⭐⭐⭐ Strongly Positive ### Reference 2: Mike Johnson [...] ## Competency Assessment | Competency | Rating | Evidence | |------------|--------|----------| | Technical | 4.7/5 | "Strongest engineer on the team" | | Leadership | 3.7/5 | "Growing into leadership role" | ## Themes **Consistent Strengths:** - [Strength 1] - [Strength 2] **Development Areas:** - [Area 1] - [Area 2] ## Red Flags [None identified / List any concerns] ## Recommendation ☑️ **STRONG HIRE** - All references enthusiastic, no concerns raised ## Raw Notes [Detailed call notes appended]
User: Run reference checks for David Kim, final candidate for Engineering Manager
Agent: I'll conduct reference checks for David Kim.
**References provided:**
1. Sarah Chen - Former director at TechCo
2. Mike Park - Peer engineer at TechCo
3. Lisa Wang - Direct report at TechCo
**Key competencies to verify:**
- Technical leadership
- People management
- Cross-functional collaboration
- Conflict resolution
**Outreach sent:** Feb 1, 2025
---
**Reference Check Complete**
## Summary
David receives strong endorsements from all three references. Former director
describes him as "top 10% of engineering managers I've worked with." Direct
report highlights his mentorship. One note: peer mentioned he can be
"too detail-oriented" in code reviews.
## Scores
- Technical Leadership: 4.8/5
- People Management: 4.5/5
- Collaboration: 4.3/5
- Conflict Resolution: 4.0/5
## Key Quotes
> "David turned around a struggling team in 6 months. Retention went from
> 40% to 90%." - Sarah Chen
> "Best manager I've had. He genuinely cared about my growth." - Lisa Wang
## Recommendation
✅ **STRONG HIRE**
No red flags. Consistent positive feedback across all references.
Minor development area (code review thoroughness) is actually a strength
for a management role.candidate-screeningoffer-letter-generator, hiring decision| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-14 | pass→pass | 10,060 | 9,848 | -2% | 1 | 1 | 0% | 1,731 | 3,352 | +94% | 0 | 0 | — |
case-01 | fail→fail | 15,165 | 15,279 | +1% | 1 | 1 | 0% | 2,609 | 4,290 | +64% | 0 | 0 | — |
case-02 | fail→pass | 15,130 | 11,944 | -21% | 1 | 1 | 0% | 2,403 | 3,728 | +55% | 0 | 0 | — |
case-03 | fail→pass | 21,104 | 13,447 | -36% | 1 | 1 | 0% | 2,456 | 3,815 | +55% | 0 | 0 | — |
case-04 | pass→pass | 15,835 | 11,358 | -28% | 1 | 1 | 0% | 2,348 | 3,590 | +53% | 0 | 0 | — |
case-05 | pass→pass | 14,169 | 17,475 | +23% | 1 | 1 | 0% | 2,318 | 3,488 | +50% | 0 | 0 | — |
case-06 | pass→pass | 12,798 | 10,252 | -20% | 1 | 1 | 0% | 2,184 | 3,342 | +53% | 0 | 0 | — |
case-07 | fail→pass | 11,939 | 9,014 | -24% | 1 | 1 | 0% | 1,833 | 3,244 | +77% | 0 | 0 | — |
case-08 | pass→pass | 9,111 | 8,066 | -11% | 1 | 1 | 0% | 1,634 | 3,039 | +86% | 0 | 0 | — |
case-13 | fail→fail | 8,627 | 5,539 | -36% | 1 | 1 | 0% | 1,266 | 2,659 | +110% | 0 | 0 | — |
case-09 | fail→pass | 9,244 | 3,805 | -59% | 1 | 1 | 0% | 1,825 | 2,416 | +32% | 0 | 0 | — |
case-10 | pass→pass | 9,447 | 3,713 | -61% | 1 | 1 | 0% | 1,498 | 2,286 | +53% | 0 | 0 | — |
case-11 | pass→pass | 9,911 | 8,540 | -14% | 1 | 1 | 0% | 1,571 | 2,966 | +89% | 0 | 0 | — |
case-12 | fail→pass | 8,049 | 4,593 | -43% | 1 | 1 | 0% | 1,283 | 2,251 | +75% | 0 | 0 | — |
case-15 | fail→pass | 3,554 | 3,166 | -11% | 1 | 1 | 0% | 704 | 2,230 | +217% | 0 | 0 | — |
case-16 | fail→pass | 5,129 | 2,517 | -51% | 1 | 1 | 0% | 807 | 2,055 | +155% | 0 | 0 | — |
case-17 | pass→pass | 22,429 | 12,980 | -42% | 1 | 1 | 0% | 2,706 | 4,053 | +50% | 0 | 0 | — |
case-18 | pass→pass | 6,178 | 3,714 | -40% | 1 | 1 | 0% | 1,010 | 2,203 | +118% | 0 | 0 | — |
case-19 | pass→pass | 5,849 | 2,937 | -50% | 1 | 1 | 0% | 940 | 2,055 | +119% | 0 | 0 | — |
case-20 | pass→pass | 8,382 | 7,377 | -12% | 1 | 1 | 0% | 1,590 | 3,010 | +89% | 0 | 0 | — |
case-21 | pass→pass | 13,651 | 9,341 | -32% | 1 | 1 | 0% | 2,322 | 3,275 | +41% | 0 | 0 | — |
case-22 | pass→pass | 19,035 | 21,556 | +13% | 1 | 1 | 0% | 3,453 | 5,673 | +64% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +32 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.