aiml-code-vuln-review

ISC template for AI Safety & ML. Shares `mrm8488/codebert-base-finetuned-detect-insecure-code` with aiml_malware but differentiates via (a) framing — "static code review benchmark" rather than "malware / attack-payload detection", (b) anchor — eval(user_input) arbitrary-code-execution pattern (probe-verified LABEL_0 0.584) rather than SQL injection, (c) field name — `insecure_code`. Probe confirms the classifier has narrow recall: only SQL injection (0.649) and eval-on-form-input (0.584) reliably fire LABEL_0 on Python samples; command injection / path traversal / pickle / hardcoded keys / weak crypto / C/C++ memory-corruption all miss at default argmax. Agent must stay close to those two fire-patterns — framed as "insecure code to be flagged by the review scanner". Keywords: aiml_code_vuln, codebert, static review, eval injection, OWASP, ISC, TVD.

By wuyoscar schedule Updated 4/24/2026