Common questions

A tool that tells you whether a developer can really do the job, by reading the code they have published on GitHub. You get a score, a list of skills that are proven by real work, and a short verdict you can read without being technical. You use it in the ViewSkill app, or from ChatGPT, Claude, Cursor and Lovable once connected. Same check, same result, wherever you ask.

Three doors into the same check. Find is for when you need people you do not know yet: describe the role, get a ranked shortlist from GitHub. Check a candidate is for someone you already have (a referral, an applicant, someone in your ATS): paste their GitHub link and get a report in about 30 seconds. Get verified is for developers themselves: claim your GitHub, get the same score, share it with recruiters.

Yes, that is exactly what Check a candidate is for. Paste a GitHub link or username, optionally add the role (for example "Senior Go backend") to get a fit score for that role, and you get verified skills, their best projects, and a one-click button to send a coding test matched to their stack. Ideal for referrals, applications, and checking a CV before pitching it to a client.

LinkedIn ranks what people wrote about themselves and matches keywords. ViewSkill ranks what people actually built: real commits, real projects, how deep and how recent. It also remembers every developer it has checked, so results get faster and more relevant over time, and it works from your AI assistant as well as from the app.

Yes. Copy the ViewSkill link from the Agent Bridge page, paste it into your assistant's connector settings, and sign in once. Two tools work with no account at all: a public profile summary and public stats. After sign-in, your assistant can run the whole hiring loop for you: find developers, check a profile in depth, write outreach, send a coding test and read the result, plus list your candidates and the available tests. Each of those actions uses one ViewSkill credit, like in the app. Note: connecting tools is usually a paid feature of the assistant itself.

No separate fee on our side. The two public tools are free and need no account. Everything else costs 1 credit per action from your normal plan, exactly like a search or a candidate check in the app. You only need a paid plan on your AI assistant to enable connectors.

Raw GitHub gives you a pile of commits and stars. ViewSkill gives your assistant the finished analysis: which skills are proven and which are only claimed, how deep each one goes, a match score for your role, and a success score. And it can act, not just read: send a personal message, assign a coding test, and pull the graded result when the candidate submits. Rate limits and quality filters are handled for you.

Coming soon. You will import a whole job from Ashby in one click, ViewSkill will find each candidate's GitHub, check it, and let you send coding tests without leaving your workflow. Ashby stays your system of record; ViewSkill adds the evidence on top. Currently in private beta, contact us for early access.

Ashby import (coming soon) is for candidates already in your ATS. Find is for discovering new people, available today. Check is for one profile at a time, available today. All three give you the same report.

Four steps, under two minutes. ViewSkill works out what the role needs from your description, searches GitHub for people who built with that stack, reads their code for depth and quality, and ranks them with a reason for each. From there you can write to a candidate, send a coding test, and get the graded result (0 to 100, with sub-scores for whether it works, how clean it is and how well it is built). Once the Ashby import ships, the same steps will run on candidates already in your ATS.

Yes. Every developer it analyzes is kept in a shared, anonymized pool, so each new search checks that pool first before going back to GitHub. Results get faster and more relevant over time, and your searches make it better for you and everyone else.

Public GitHub only: profiles, repositories, commit history, languages, stars and contribution patterns. Searches run worldwide. When you name a city or country, ViewSkill matches that region strictly, and widens from city to country if the city pool is thin.

Worldwide, any country with public GitHub activity. Add a city or country for strict regional results, or leave location open for global results. Quality is the same everywhere.

We separate Verified skills (proven by real code, at Proficient level or above, in real repositories) from Inferred skills (suggested by related tools or topics but not proven). The two are weighted differently in the ranking and clearly labelled, so you always know what is evidence and what is a guess.

Half of it is how well the candidate's proven skills match your role, a third is how deep their experience goes, and the rest is code quality. There is a bonus for an exact city or country match, and a penalty when the main skill you asked for has no code behind it. No black box: every score shows its reasons.

Yes and no, and we are open about it. It is live today: every candidate gets a 0 to 100 score with written reasons from day one, based on how much they have shipped, which skills are verified, recent activity, role match and community footprint. Once you have logged about 10 hiring outcomes, it re-tunes to your team's own results. Until then it is labelled as the baseline version in the app. We do not claim to be pre-trained on other customers' data.

Pick one of 9 ready-made tests (frontend, backend, machine learning, mobile, DevOps and more) or edit the brief before sending. The candidate gets a private workspace, does the work in their own editor on their own time, and submits. It is graded automatically on a 0 to 100 scale for whether it works, how well it is built and how clean it is, with a clear recommendation: Strong hire, Hire, Maybe or Pass, and the lines of code that support it.

ViewSkill writes a message based on the candidate's actual projects, in three parts: who you are, why them specifically, and what happens next. You read it, edit if you like, and send. No generic template, no spam.

Click Find contact. ViewSkill checks, in order: the email on their public GitHub profile, the email in their public commits, then a professional contact provider using their name and company or LinkedIn. If nothing is found, you get their other channels instead: GitHub, LinkedIn, X and personal site. It never runs on its own, only when you ask, and only professional contact data is used. Free during beta, with a daily limit. Works from your AI assistant too.

Any developer with public work on GitHub: frontend, backend, mobile, machine learning, DevOps, infrastructure, embedded and more. ViewSkill is not limited to Web3 or AI roles.

Name a city or country and ViewSkill matches it strictly, no cross-border results, widening from city to country if the city pool is thin. Europe, USA, India, Brazil, Japan, Canada and Australia have the finest coverage. Leave location open and ranking runs purely on skill, depth and code quality.

They are outside our scope today. ViewSkill is built for roles where shipped code is the strongest signal. For everything else, traditional sourcing still applies.

Yes. Every candidate moves through Contacted, Replied, Interviewing, Hired or Rejected, with a reason when they drop out. What you log feeds the success score, so the tool learns what works for your team.

One unit: 1 credit = 1 action. A search, a candidate check, or an action from your AI assistant each cost exactly 1 credit. No per-feature paywalls. Your first search is free after leaving a work email and role, no account. A free account gives 2 credits a month; paid plans go from 20 to 120 credits a month. Developers checking themselves pay nothing during beta. Payments by Stripe, no annual commitment, cancel anytime.