Research — Prsho

We publish what our tools do and how well — not the internal details of how.

What we work on

• Correctness before coverage. A report should rather say “not established” than state something untrue. We measure how well our tools hold to this.
• Threat model: the repository is hostile. How to analyze code you do not trust without running it.
• Measurement protocol. Expectations are written down before each run, and the bar for success is fixed in advance and not moved afterwards.

Results

No results are published yet. We will publish measured results here once the final measurement is complete — including where the tools fall short.