LegalPrivacy Policy
Last updated · October 22, 2026Privacy is a fundamental right, not a luxury. Pragmys is built on a minimal data footprint — we don't collect, harvest, or sell your personal information. Your code remains yours, and your profile is built entirely on your own terms.
Overview
Our philosophy
At Pragmys, privacy is a fundamental right, not a luxury. Our platform is built on a minimal data footprint. We do not participate in the data economy — we do not collect, harvest, or sell your personal information. Our systems prove that verified engineering credentials and un-fakeable proof-of-work can be generated without compromising digital anonymity or proprietary source code.
Intentional privacy & source code protection
Our platform is privacy-first by design. Pragmys evaluates the actual projects you build to generate an engineering skill graph, but we never store your raw source code.
When we analyze your repositories (whether via local CLI or GitHub integration), the raw source code is discarded immediately as soon as the structural analysis is finished. We only retain the derived insights — such as architectural patterns, your verified skill graph, and telemetry footprints — required to maintain your Pragmys profile. Your private code never sits on our servers.
Your data
Data collection & usage
To provide the Pragmys experience and maintain your verified profile, we process only the data strictly necessary for functionality:
Secure authentication: We use industry-standard protocols to secure your account, storing only the identifiers necessary to manage your session.
Behavioral & structural insights: To facilitate accurate skill mapping, we process the structural skeleton of your repositories and aggregated telemetry, enabling personalized project recommendations and accurate role-readiness diagnostics.
AI usage & intent telemetry: To calculate your AI orchestration skills and verify authentic code ownership, our local systems evaluate your AI conversations (such as IDE prompts) and localized code snippets. Crucially, all AI conversations and snippets are strictly redacted and PII-scrubbed locally on your machine. Only the sanitized, anonymized telemetry is securely transmitted for analysis.
Zero-PII execution: All code redaction and PII-scrubbing happens safely on your machine before any telemetry reaches our cloud engine.
Data sovereignty & visibility control
Your data exists only to serve your growth and career trajectory. We implement robust security layers to ensure your information is protected. Because we avoid storing raw source code, your digital safety is inherent to our minimalist architecture. You maintain full control over who sees your profile:
Public links: Your data is shareable using public portfolio links strictly based on the visibility preferences you set in Pragmys. You choose what to reveal.
Opt-in job matching: To connect your verified profile with active career pipelines, your insights can be shared with hiring companies for job matching — but only with your explicit, proactive consent.
Deleting your data
You can delete your profile at any time from Profile → Delete my profile. It takes two confirmations and cannot be undone.
You can also download a copy of your data at any time from Profile → Download my data. It is a single JSON file and leaves out sign-in credentials and share-link tokens.
Deleted: your account, public page and share links, repository analysis and skills, AI reports, Qualify plans, follows, sign-in sessions, connected Compass agents and their run records, plus the stored evidence packages. We also remove the push webhooks we added to your repositories and revoke our GitHub authorization.
Kept, anonymised: job-check statistics. The posting's role title, platform, skill names, counts and dates stay as aggregate analytics, with no link to you and none of the written feedback. A job title can name an employer, so we never publish these rows.
Not removed: repositories or workflows Pragmys created in your own GitHub account while scaffolding a project. They are yours; delete them on GitHub if you no longer want them.
Sharing
Third-party transparency
Pragmys as an organization does not share your performance, insights, or account data with any third party for advertising or marketing purposes. We only interact with external services essential for secure account management, structural AI evaluation, and platform stability.
Analytics
We partner with Microsoft Clarity to capture how you use and interact with our website through behavioral metrics, heatmaps, and session replay to improve our products/services. Website usage data is captured using first and third-party cookies and other tracking technologies to determine the popularity of products/services and online activity. Additionally, we use this information for site optimization and fraud/security purposes. For more information about how Microsoft collects and uses your data, visit the Microsoft Privacy Statement.
Products
Compass Browser: extension & permissions
The Compass Browser extension (part of Compass by Pragmys) has a single purpose: it shows you how well your verified Pragmys skill profile matches a job posting you're viewing, directly on supported job sites (LinkedIn, Naukri, Indeed, Glassdoor, Wellfound, and Unstop).
The extension scores a posting in two tiers, and what leaves your device is different for each:
Instant tier: The moment you extract a job posting, it's scored entirely on your device against a cached copy of the public skills taxonomy and your own cached skill confidences. Nothing about the job posting is sent anywhere for this tier.
Deep analysis tier: Only when you explicitly click for the full breakdown, the job title and description are sent to api.pragmys.com so our LLM can extract the specific skills the posting requires and generate a written match summary.
What we retain from a deep analysis: the derived result — matched, weak, and missing skills, your match score, and the written summary — plus the job title and the domain of the site it came from (e.g. "linkedin.com"), so the analysis appears in your diagnostics history at pragmys.com. We do not retain the full text of the job description or the specific URL of the posting. That content is used only to compute the match for your request and is not written to our database.
It requests the following browser permissions, each used strictly for that purpose:
activeTab: Used when you click the extension icon or the in-page widget to read the currently open job posting's title and description so we can score it against your skills. The extension never reads a tab you haven't actively interacted with.
scripting: Used to inject a content script that extracts job title and description text from the active job-posting tab, and as a fallback extractor on pages where the primary content script hasn't loaded yet.
storage: Used to cache your auth session token, your skill profile, and the public skills taxonomy locally in your browser, so the extension can score jobs instantly without re-fetching data on every page view.
alarms: Used to periodically check for completion of the device-code login flow you initiate from the popup, without keeping the background service worker running continuously.
cookies (optional): Requested only if needed, to check for an existing pragmys.com session cookie so you don't have to log in again inside the extension if you're already signed in on the web.
Host permissions: Access to job-board domains (LinkedIn, Indeed, Glassdoor, Naukri, Wellfound, Unstop) is required to inject the in-page match widget and read job content on those specific sites. Access to api.pragmys.com is required to send the job title and description to our backend for the deep-analysis tier described above.
The extension does not download or execute remote code — all logic ships inside the packaged extension bundle, and its network requests only exchange data (job text in, match result out), never executable code. Consistent with the rest of this policy, the extension does not sell your data or share it with third parties for advertising or marketing.
Compass (MCP server)
Compass is a local MCP server that runs on your device with your own AI agent. Your agent finds job postings; Compass scores them against your verified skills. This is what happens to data when you use it.
Postings stay on your device. Job postings, feeds and the HTML report stay on your device and with your own agent and its model provider, under their terms. Compass does not send Pragmys servers any posting text, and Pragmys does not fetch job sites on Compass's behalf. Radar, a separate feature, reads public employer career boards; see Radar below.
Agent sign-in: When you connect an agent, Pragmys stores a hash of its token with the label shown on your profile, its expiry and when it was last used. The token itself stays in your operating system's keychain, or in a file only you can read. You can revoke any agent under Profile → Connected agents.
Your skills: When your agent asks for evidence, Pragmys returns your skill names and confidence scores and nothing else, within hourly and daily read limits.
One usage record per run: the role you searched, whether you gave a location and how many terms, how many deep checks you asked for and finished, how many postings were scored, whether a report was written, how long the run took, the Compass version, and counts, timings, refusals and failures per source. It never includes posting text, links, company names, locations, results or skills. Run compass_usage_preview to see the exact record, or set COMPASS_USAGE=0 to send none.
Website "Find jobs" card: when you are signed in, Pragmys records that you saw the Compass card on Career or Dashboard and whether you clicked an install or Browser link or copied a prompt: the event name, which page and the time, nothing else. It is removed after 90 days and when you delete your profile.
How long we keep it: the role is removed from a run record after 90 days and the counts stay. The counters behind the read limits are removed after two days. Deleting your profile removes your agent tokens, sign-in codes and run records.
The full details, including install steps and what Compass does not do, are on the Compass page.
Radar (job search)
Radar is the signed-in job search inside Pragmys. It ranks open roles against your verified skills. This is what happens to data when you use it. The rules for using it are in the Terms of Service.
Where the postings come from. Pragmys servers read public career boards that employers publish through open applicant-tracking systems (such as Greenhouse, Lever, Ashby and Workable). We keep the job title, company, location, structured facts (level, workplace, salary where stated), the skills we detect, and a short excerpt. We do not keep or show the full posting. Every role links to the employer's own page to apply. The postings belong to the companies; Pragmys does not apply to jobs for you.
How we rank for you. Your skill names and confidence scores are compared with each posting's detected skills, on our servers. Radar sends no resume, repository content or source code anywhere. For your top roles, an AI model reads the posting text, not your profile, to extract the skills it asks for; your fit is then computed from that result on our side. Personalised recommendations run on Pragmys' servers and are used only to personalise your choices, not for any commercial benefit.
What Radar stores about you. Roles you save, apply to, ignore or add notes to, with a timeline of status changes and a small snapshot of each (title, company, link, excerpt, fit) so your history survives after the posting closes. Searches you save as alerts. Companies you follow. Alerts appear in the app only; Radar does not email you.
Usage data. To improve ranking and to check that match scores point to useful roles, we record how you use Radar: results shown with their position and score, opens, saves, apply clicks, filters and searches. This data is kept in a separate database that holds no name, email or profile details, and is linked only by an internal account identifier. It is kept for 180 days; only aggregated counts are kept longer. It is never shared with employers or third parties and is not used for advertising.
Apply links. Apply buttons go through a Pragmys link so we can remember that you clicked and later ask whether you applied. The employer receives nothing from us; their site sees an ordinary visit.
Public page. Signed-out visitors see only market totals (roles, top skills, domains, median salaries). Every figure covers many postings and is rounded, and no employer or role is named.
How long we keep it. Daily posting history and closed postings: 30 days. Raw usage data: 180 days. Tracker rows you ignored: 180 days after the posting closes. Everything else stays until you delete it.
Deleting your data. Profile → Delete my profile also removes your Radar tracker, notes and timeline, alerts, followed companies, usage data and cached feed. The tracker page can export your tracker as a CSV.
Takedowns. If you represent a company and want a posting or your company's information removed, or you see something wrong, email
hello@pragmys.com with the company name and the address of the board or posting. We take the specific item down within one week.
About
Disclaimer
Pragmys is an independent software development entity and is not associated with, or a subsidiary of, any other corporation.