Facts about DevGhost

What DevGhost is

• DevGhost independently assesses each developer’s productivity from the content and cognitive complexity of Git changes, for business owners, CEOs, CTOs, and engineering leaders.
• It analyzes Git history from GitHub and GitLab, including public repositories and repositories an authorized user is allowed to connect.
• It estimates the cognitive difficulty of code changes. It does not treat lines changed, commit count or time at a keyboard as engineering output.
• Estimated effort is expressed as the hours a reference developer would need to understand the relevant code, design, implement, self-test, and correct the change.

What Ghost% means

• Ghost% compares individual code output with a baseline of 3 reference hours per developer-day, adjusted for the selected Spread or FTE day count and Coding Share. Team summaries weight individual percentages by effort.
• The reference is a mid-level developer with three to four years of experience who knows the codebase and works without AI assistance.
• A sustained result above 100% means the code changes represent more estimated work than the reference baseline. Modern AI tools can produce that leverage, but a high Ghost% alone does not prove that AI was used.
• Ghost% is intended to surface patterns and questions for investigation. It is not a direct measure of code quality, business value or all work performed outside the repository.

Method and privacy

• DevGhost combines machine learning and its own neural network with structural analysis of Git changes and specialized algorithms. DevGhost does not use LLMs to analyze code.
• DevGhost uses repository history. It does not use screen capture, keystroke logging or employee time tracking.
• The estimate does not see meetings, mentoring, architecture work, product work or other contributions that leave no trace in the analyzed code.
• The service is self-service: a user can create an account, connect GitHub or GitLab, select a repository and run an analysis in the production product.

Responsible use

• Use DevGhost to identify sustained low contribution and strong developers, with independent human assessment and role context. Do not use it as an automatic KPI, public ranking or the sole or primary basis for a personnel decision.
• A low or high result should be reviewed across more than one period and interpreted with the developer, team lead or CTO who knows the context.
• Different responsibilities, blockers, part-time allocation and work outside the repository can explain an apparent deviation.
• Useful applications include finding delivery bottlenecks, validating concerns, recognizing hidden leaders, comparing internal and external delivery capacity and studying the effect of AI adoption.

Company and founder

DevGhost was founded in 2026 by Pavel Kosyakov, a software engineering and product leader with more than 20 years of industry experience. Before DevGhost, he led software development, developer experience and engineering analytics initiatives at Sber and Sberbank-Technologies.

LinkedInGitHub