Skip to main content
Evidence items in evidence.json are links to external proof of a candidate’s skills: GitHub repositories, certifications, portfolio sites, publications, and more. Verifying these links before an interview gives you a trust score for the candidate’s claimed credentials.

​
Verifying evidence

The verification function performs an HTTP HEAD request on each evidence URL and checks whether it is reachable. 
def verify_evidence(evidence_json):
    """Verify each evidence item is reachable."""
    results = []
    for item in evidence_json.get('items', []):
        url = item.get('url', '')
        try:
            resp = requests.head(url, timeout=5, allow_redirects=True)
            results.append({
                'title': item['title'],
                'url': url,
                'verified': resp.status_code < 400,
                'status': resp.status_code
            })
        except:
            results.append({
                'title': item['title'],
                'url': url,
                'verified': False,
                'status': 'unreachable'
            })
    
    verified = sum(1 for r in results if r['verified'])
    return {
        'total': len(results),
        'verified': verified,
        'trust_ratio': verified / max(len(results), 1),
        'items': results
    }
Requests time out after 5 seconds. URLs that fail to respond within this window are marked as unreachable and counted as unverified. Adjust the timeout if your network environment has higher latency.

​
Trust levels

The trust_ratio returned by verify_evidence maps to five trust levels:
Trust ratioLevelInterpretation
1.0 (5/5)Level 4 — Full trustAll evidence verified
0.8 (4/5)Level 3 — High trustOne minor dead link; proceed
0.6 (3/5)Level 2 — Moderate trustReview unverified items manually
0.4 (2/5)Level 1 — Low trustSignificant gaps; ask candidate to update
0.0–0.2Level 0 — No trustMost links unreachable; treat as unverified

​
Handling verification results

Verified evidence (verified: true) can be used to confirm the candidate’s claimed skills. Cross-reference the evidence type against the skills listed in profile.json. Unverified evidence (verified: false or status: 'unreachable') should not count against the candidate immediately — links go dead for reasons outside their control. Flag these items and ask the candidate to update them if possible.
A trust ratio of 0.8 or above (4 out of 5 items verified) is generally sufficient to proceed. A single dead certification link does not disqualify an otherwise strong candidate.