[HN Gopher] A security scanner as fast as a linter - written in ...
___________________________________________________________________
A security scanner as fast as a linter - written in Rust
Author : peaktwilight
Score : 31 points
Date : 2026-04-08 14:05 UTC (2 days ago)
(HTM) web link (github.com)
(TXT) w3m dump (github.com)
| mplanchard wrote:
| Looks interesting, will give it a run on the codebase at $work.
| One thing that would be nice to see in the README are benchmarks
| on larger codebases. Everything in the benchmark table is quite
| small. I'd also list line count over files, since the latter is a
| much better measure of amount of code.
|
| For context, the codebase I work on most often has 1200 JS/TS
| files, 685 rust files, and a bunch more. LoC is 13k JS, 80k TS,
| and 155k Rust
| mplanchard wrote:
| It is still quite fast on that codebase, fwiw. 10.7 ms.
| staticassertion wrote:
| Legitimately, I have had to stay away from certain linting tools
| because of how slow they are. I'll check this out.
|
| cfn-lint is due for one of these rewrites, it's excruciating. I
| made some patches to experiment with it and it could be a _lot_
| faster.
| woodruffw wrote:
| Some of the checks here seem very brittle. For example this
| one[1].
|
| In the context of security scanning (versus, say, listing), I
| think it's reasonable to expect the tool to be resilient to
| attempts at obfuscation (or just badly written code that doesn't
| adhere to normal Python idioms around import paths).
|
| [1]: https://github.com/PwnKit-
| Labs/foxguard/blob/a215faf52dcff56...
| yatac42 wrote:
| From a quick look it seems like it's "as fast as a linter"
| because it _is_ a linter. The homepage says "Not just generic
| AST patterns", but I couldn't find any rule that did anything
| besides AST matching. I don't see anything in the code that would
| enable any kind of control or data flow analysis.
| kabir_daki wrote:
| Running security checks at linter speed is a big deal for CI
| pipelines. What's the false positive rate in practice? That's
| usually the tradeoff with fast static analysis -- speed vs
| accuracy. Would love to know how you benchmarked it.
| davewritescode wrote:
| The speed is really cool but the fact that your rules are written
| as rust code meaning that new rules need a new binary. That might
| be fine but just wanted to point it out to anyone who's
| interested.
___________________________________________________________________
(page generated 2026-04-10 23:00 UTC)