Trust & Compliance
Every hiring decision, ready for scrutiny
Automated hiring tools are under growing legal and regulatory scrutiny — and they should be. SkillFoundry is built so every assessment decision can be explained, audited, and defended.
Compliance posture
A primary blue dot indicates a capability that is live today; amber indicates an active workstream with a target date. We describe our status plainly — see the roadmap note below for what we have not yet completed.
Evidence and Auditability
Every score on SkillFoundry is assembled from recorded evidence rather than a reviewer's recollection:
- Trusted test execution. Submissions are graded by running the task's tests on the code the candidate actually submitted, in a trusted environment, and the results are stored with the submission.
- Code quality review. The submitted code is reviewed for quality, and the result appears in the score breakdown.
- Session event records. Assessment sessions are captured as structured behavioral interaction events, which inform the behavioral portion of the result and provide a trail reviewers can inspect.
- Evidence-based assessment. Results describe how a candidate investigated, used AI and verified their work, and reviewers can verify each result against the evidence behind it. Missing evidence is never counted against a candidate. SkillFoundry is live. Evidence-based results become the primary score after a readiness review.
- Persistent score breakdowns. Each evaluated submission stores its full breakdown — base score, test results, code-quality metrics, behavioral scores, AI-use assessment, and every adjustment applied.
Human Oversight, On the Record
Automated scores on SkillFoundry are recommendations, not verdicts. The platform is designed for human-in-the-loop review:
- Reviewers can adjust any evaluated score through a structured manual-review workflow.
- Every manual adjustment is logged with the reviewer's identity, the change, the stated reason, and a timestamp — and the log is preserved on the submission record, and the original result is never overwritten.
- Organizations set their own assessment policies, including whether candidates may use AI assistance. Score adjustments related to AI use follow the organization's configured policy and are disclosed in the report.
Responsible AI Use, Not AI Detection
We do not claim to catch every possible use of AI, and we caution customers against vendors who do. SkillFoundry's approach is transparency over certainty:
- Using AI is not penalized in itself. Reports describe the judgment a candidate applied to AI assistance rather than issuing a binary "cheating" verdict.
- What the platform cannot observe is reported as such; it is never guessed or counted against the candidate.
- Findings are presented to reviewers as evidence, so a human makes the final call, with the reasoning recorded.
Data Protection
- Transparent monitoring. Candidates are informed about what the platform collects during an assessment session before they begin. Our monitoring and data-collection practices are documented publicly.
- Retention and erasure. Candidate data is retained under our published Privacy Policy. On a verified request, behavioral assessment evidence is deleted from live systems within 48 hours, subject to legal holds; other account data is deleted or anonymized in line with our retention policy. Backups are generally kept for up to 90 days and roll off; if a backup is ever restored for disaster recovery, pending deletions are applied again.
- Data processing agreements. A DPA covering our role as a processor of candidate data is available to customers on request.
- Access controls. Assessment data is scoped to your organization; reviewer actions require authenticated, role-based access.
Behavioral Evidence: What We Keep and For How Long
Behavioral events, not keystrokes
We record behavioral interaction events during an assessment, such as workspace activity, test runs and use of the AI assistant. Never keystroke content, webcam, microphone or screen recording.
Code and transcripts stay protected
Code changes, test reports and AI assistant transcripts are stored as protected artifacts. Secrets are redacted, only authorized reviewers can open them, and access is logged.
24 months for evidence, 12 for artifacts
By default, behavioral evidence is kept for 24 months and protected artifacts for 12 months. Organizations can set shorter or longer periods within fixed bounds.
Evidence deleted within 48 hours
On request, behavioral assessment evidence is deleted from our live systems within 48 hours, unless a legal hold applies. Other account data is deleted or anonymized in line with our retention policy. Backups are generally kept for up to 90 days and roll off; if a backup is ever restored, pending deletions are applied again.
Research use only with opt-in consent
Practice attempts are used to improve our analysis only if the candidate explicitly opts in. Consent is off by default, separate from the terms, limited to people 18 or older, and can be withdrawn in Settings at any time. Organization assessments are used for calibration only where the organization has agreed to it by contract.
Pseudonymized research copies
Research copies are pseudonymized: names, emails, code and transcripts are removed. They are never used to score anyone and never shown to employers.
Transparency for Candidates
- Candidates are told what is captured before a session, and the workspace shows when activity is being recorded.
- Where evidence-based assessment is the primary result, candidates receive descriptive, plain-language feedback.
- Using the built-in AI assistant is allowed wherever the organization permits it.
More about evidence-based assessment, or talk to us.
Fairness and Automated-Hiring Regulation
Laws such as New York City's Local Law 144 place obligations on employers who use automated employment decision tools, including bias audits and candidate notice. SkillFoundry is designed to make meeting those obligations practical:
- Job-related inputs only. Scoring operates on work products and session behavior — code, tests, and how the work was done. Demographic attributes are never inputs to any score.
- Consistent criteria. The same rubric and scoring model apply to every candidate on a task, eliminating reviewer-to-reviewer variance in the automated portion of the evaluation. Each attempt is pinned to the scoring profile version it was scored with, and task packs are reviewed by a named human before publication.
- Audit-ready records. Per-candidate score breakdowns, session evidence, and manual-review logs give your compliance team the raw material an independent bias audit requires.
- Human review built in. Structured manual review keeps a person accountable for every final decision.
Where we are honest about the roadmap: SkillFoundry has not yet completed a third-party SOC 2 examination or published an independent adverse-impact study, and we will not imply otherwise. Both are active workstreams; customers evaluating us can contact our team for current status, our security documentation, and interim attestations.
Questions from your security or legal team?
Tell us you need a security or compliance review and we'll follow up with data flows, the evidence model, and our compliance posture.