Every other tool sorts developers by their own description of themselves. ViewSkill reads the code they published, rates it, and tells you in plain English who is worth your time.
LinkedIn Recruiter, job boards, GitHub search
- Rank people by what they wrote about themselves
- Cannot tell an expert from someone who took a course
- Same shortlist for every recruiter who types the same words
- Never learn who you actually hired, or how it went
- Screening still needs an engineer's afternoon
ViewSkill
- Rates people by the code they published, not their CV
- Rates every skill from Familiar to Expert, with the proof one click away
- Predicts who will succeed in your role, and gets sharper with every hire you log
- Checks candidates you already have: applicants, referrals, your ATS
- Writes to candidates, sends coding tests and grades them for you
- Works inside ChatGPT and Claude if that is where you work
- One price per action, no seats
Knowing someone can code is not the same as knowing they will succeed.
Most tools stop at the first question. Recruiters get paid on the third.
- 1
Can they code?
“Does this developer know React?”
A yes or no. Useful, and soon every tool will offer it.
- 2
How well, and how recently?
“How deep is it, and are they still doing it?”
Five levels from Familiar to Expert, based on how many projects, how much recognition, and commits in the last six months. Separates people who tried it from people who ship it.
- 3, what ViewSkill is built for
Will they succeed with you?
“Will this person still be performing at six months, in your team, by your standards?”
A score today, tuned to your own outcomes over time. After about 10 logged hires it learns what success looks like for you.
Every candidate gets a success score from day one. It is explainable, so you can show the hiring manager why. Once you have logged about 10 hires, it re-tunes to your team's definition of success.
Sample scores.
We don't trust labels. We read the code.
A skill tag on a profile weighs the same whether the person built one tutorial project or runs a library used by thousands. ViewSkill tells the difference, even when they never tagged their work.
On a CV or LinkedIn
Self-declared. No proof, no depth.
ReactTypeScriptNodeDockerAWSGraphQL
Same weight for every tag. You cannot tell expert from beginner.
On ViewSkill
Rated on real projects, with the evidence attached.
- React
- Advanced, verified
- TypeScript
- Proficient, verified
- Node
- Proficient, verified
- Docker
- Working, transferable
- AWS
- Mentioned only
8 React projects, 240 stars. Verified by code, not by claims.
Five levels, from Familiar to Expert
Every skill is rated on three things anyone can understand: how many projects use it, how much recognition those projects got, and how recently the person worked with it. Proficient and above earns the Verified label.
| Level | Meaning | Repos | Stars | Recent activity |
|---|---|---|---|---|
| L1 | Familiar | 1 repo | <5 stars | occasional |
| L2 | Working | 2-3 repos | 5-20 stars | this year |
| L3 | ProficientVerified | 4-6 repos | 20-100 stars | last 6 months |
| L4 | AdvancedVerified | 7-12 repos | 100-500 stars | this month |
| L5 | ExpertVerified | 12+ repos | 500+ stars | this week |
- What we look at
- We scan repository languages, names, descriptions and commit recency to detect real expertise, not just self-declared topics.
- How the level is set
- Every skill gets an objective depth score from Familiar to Expert, based on number of repos, stars and recent activity.
- Verified or not
- Skills with strong repo evidence (Proficient and above) are promoted to Verified. Weaker signals stay clearly labeled as transferable or inferred, so you always know what's solid.
See where candidates drop off, and why
Every candidate moves from found to contacted, interviewing, offer, hired. Log the reason when someone drops out and the patterns that hurt your hiring show up on their own.
Sample funnel. Your real numbers appear in your dashboard.
Built for the people who own the hire
Agencies, in-house teams and the managers who sign off. One bar across every channel.
Recruiting agencies
Stop pitching candidates you have not checked.
Paste any candidate's GitHub link and send your client a report backed by real code, today. Soon, import a role from Ashby in one click and get a report for every applicant in minutes.
More of your pitches turn into interviews.
In-house talent teams
One bar for every channel.
Referrals, applicants, agency candidates and your ATS all go through the same check. No inflated CVs, no favourites, one consistent technical standard you can defend.
Hiring decisions you can explain to the hiring manager.
Hiring managers and CTOs
Skip the one-hour technical screen.
Read the report on their real work, then send a coding test matched to their stack in one click. Spend engineering hours only on people who already passed the evidence gate.
Get 5 to 10 engineering hours back per hire.
Work inside ChatGPT, Claude, Cursor or Lovable? The same checks run there too. See how
Stop guessing. Start checking.
Try it on a real role or a real candidate. No credit card.