Getting started for hiring teams¶
This guide is for interviewers and recruiters: how to get access, create assessment tasks, invite candidates, and review results.
1. Get access¶
Interviewer accounts belong to an organization. There are two ways in:
- Accept an invitation — your organization admin sends you an invite with the Interviewer role. Open the link, sign up (or sign in), and accept.
- Self-register — sign up, choose Interviewer at role selection, and enter your organization's ID when prompted. If you don't have the organization ID, ask your admin or have them invite you instead — you cannot proceed as an interviewer without an organization.
Once in, your home is the Interviewer Dashboard, which lists your organization's tasks with status filters and actions to create or import new ones.
2. Create assessment tasks¶
You have several ways to create tasks, in increasing order of effort:
| Method | Best for | Where |
|---|---|---|
| Use a public template | Getting started quickly | Templates catalog — filter by difficulty and language |
| Import a pull request | Assessing on your real codebase | Import PR — paste a PR URL; SkillFoundry snapshots the pre-PR repository state and generates a task from the change |
| Task Builder | Polished, branded tasks from an issue + PR pair | Task Builder — provide an issue URL and PR URL; the builder drafts a candidate-facing brief you can edit before publishing |
| Create a template from scratch | Fully custom challenges | Templates → Create |
Key task properties:
- Visibility — private to your organization, or public/community.
- Difficulty and tags — used for filtering and matching.
- AI policy — whether candidates may use AI assistance on this task. The organization also has a default AI policy; the task setting and the organization setting both apply, so make sure they agree with what you tell candidates.
- Time limit — optional; the countdown starts when the candidate starts the attempt.
Full details in Creating and managing tasks.
3. Invite candidates¶
From Candidates in the interviewer area:
- Click Invite candidate and enter their email.
- Optionally attach a task assignment to the invitation, so the task is waiting on their dashboard when they accept.
- The candidate receives an email link. Invitations expire, and you can resend from the invitations list.
Note that inviting candidates counts against your subscription plan's candidate limit — if invites start failing with an upgrade prompt, see Billing and subscriptions.
4. Candidates complete the assessment¶
Candidates work in the browser code workspace or, for supported tasks, in VS Code with the SkillFoundry extension. Either way you get the same artifacts: the submitted code (as a diff), test results, and behavioral telemetry. You can point candidates at their getting-started guide.
5. Review submissions¶
The Evaluations page is your review queue, with submissions filtered by status (pending / reviewing / reviewed). For each submission you can see:
- Automated evaluation — test results, code quality analysis, and the composite score.
- Score report — overall score plus a breakdown across correctness, code quality, test coverage, working process, and communication, with correctness weighted most heavily.
- AI-usage flags — if the task disallowed AI and the session's activity looked inconsistent with that policy, the submission is flagged for your review. A flag is a prompt to investigate, not an automatic verdict — see Reviewing candidates.
- Manual review — you can adjust points with a documented reason and add feedback.
Reports gives you aggregate views: completion reports per task and per candidate.
6. Understand behavioral insights¶
Beyond the code itself, SkillFoundry evaluates how candidates work: how they approached the problem, how they validated their work, how they handled setbacks, and how they communicated. Results are presented with the supporting evidence so you can check them against the code, plus red flags for serious issues (for example, committing secrets).
For what this means for candidate privacy — and what you can tell candidates is and isn't collected — see Monitoring and data collection.
Common pitfalls¶
- AI policy confusion is the #1 source of candidate disputes. Decide your policy up front, set it consistently at the org and task level, and state it in your candidate communication.
- Expired invitations and attempts — invitations and attempts both have expiry. Check the invitations list before assuming a candidate ignored you.
- Paste-heavy candidates — legitimate workflows (copying from docs, moving code between files) can look unusual in activity data. Flags need human judgment.