Type Theory With Erasure

The feed contains multiple PL research papers; two items have direct security implications. (1) “Symbolic Execution Meets Multi-LLM Orchestration” presents a 4-agent LLM architecture that synthesizes KLEE-compatible harnesses for incomplete Rust CVE snippets (structs, imports, manifests missing), enabling symbolic analysis where existing tools fail. On 31 real-world Rust CVEs the system achieved 90.3% wrapper compilation success (vs. 0% for prior formal tools), detected 1,206 critical errors across 26 files (83.9% detection rate) and outperformed Clippy and Miri. (2) “Semia” is a static audit/