Sink boundary: Input reaches a dynamic code boundary where text can become executable instructions.
The pack does not provide a callable signature for this record; the watched access path is shown directly.
| Arg | Access path | Purpose | Watched |
|---|---|---|---|
| Argument[1] | Argument[1] | The access path Atropos marks for this model. | ▲ sink |
Attacker's-eye view · code-injection
An attacker supplies syntax that is evaluated with the process's privileges.
// modeled boundary: Argument[1]
dlmopen(attacker_influenced_value);
1Keep code and data separate.
2Replace evaluation with a parser or allow-list.
3Constrain the execution context if evaluation remains.
These checks require the surrounding codebase. The model names the boundary; it does not decide reachability or prove that a guard dominates every path.
Is this a bug in your code?Atropos stops here — by design
Atropos identifies Argument[1] as a code-injection sink. It cannot see whether untrusted data reaches this call in your repository.
Lachesis is the codebase-level step: it traces reachability and guards for this symbol.
Check this symbol in Lachesis →Same behavior
●dlopenc●LoadLibraryAc●compilejavascript●evaljavascript●evaljavascript●evalClosurejavascriptAcross languages
No cross-language match.
What neutralizes it
●literal_evalpythonModel facts · verbatim from the pack
| Role | Kind | Access path | Model ID | Confidence |
|---|---|---|---|---|
| sink | code-injection | Argument[1] | c.std.dlmopen.a1 | medium · corrob. 1 |