Roadmap

Shipped, next, and planned — kept apart

Everything under Shipped is in the product today, with the commit that shipped it. Everything else is not available yet. Dates for planned work are not promises.

Shipped

2026-10-30e2f90eaC# proofs from the batch's own call sites; Java request wrappers across files; Rust integer-overflow and unsafe-memory checks; C/C++ allocation sizes and loop bounds; Go, PHP, Ruby, Kotlin, Express and Android rules
2026-10-29bbe15ddC# proof pass and network sources; PHP, Android, DOM and Juice Shop coverage
2026-10-29043a098Java proof pass; Rust shape and log sinks; relative-URL guards
2026-10-29a21ecb3C/C++ data-flow and memory-safety engine
2026-10-29c71997fConstant folding and log injection in every engine
2026-10-27b53029ePrecision and recall work from NodeGoat, Juliet (Java/C#) and OWASP BenchmarkJava
2026-10-275fff17fGo, Rust, PHP, Ruby, C# and Kotlin precision and recall from the sast-benchmark corpus
2026-10-24b73e12fPython engine on OWASP BenchmarkPython (score 12 → 95)
2026-10-24c018d07Large pull requests fetched as one archive; scans fail closed on unread files
2026-10-23358eba2Merge gate follows the organisation's Block CRITICAL / Block HIGH settings
2026-10-21b3ea10cSign-offs follow file content across scans

Release notes: changelog. Measured effect: benchmark history.

Working on next

Follow return values across files in the C# proof pass

The largest remaining source of C# SQL false positives (Juliet flow variant 61).

Resolve settings read from .properties files in Java

Hash and cipher algorithms chosen in configuration are currently missed (BenchmarkJava hash category).

A labelled corpus for trace accuracy

So the report can measure whether each trace names the right source and sink, not only whether a finding exists.

CodeQL and Semgrep run under identical conditions

To publish a fair comparison — wins and losses — on the same corpora.

Planned

Kotlin scored on a public labelled corpus

Kotlin is tested by unit tests and deliberately vulnerable apps today, but no public labelled corpus exists to score it on.

Ruby and Rust authorization and CSRF categories

Not detected yet on the sast-benchmark corpus.

Independent penetration test of TrustLedger itself

We will publish a summary on the Trust Center when it is done.

SOC 2 audit of TrustLedger

Not started; the Trust Center will say when it is.