New pipeline recovers source code from stripped binaries using anchor-based retrieval and LLM reasoning
Researchers propose a method to recover source code from stripped binary functions by combining reverse engineering with anchor-based retrieval and LLM reasoning. The pipeline extracts anchors (strings, constants, external calls) via Ghidra, searches a source code database, and uses an LLM to re-rank candidates. This approach aims to identify exact source functions rather than generating approximate decompiled pseudocode.
Entities
Related
LVRP: Open-source local vulnerability research pipeline uses 14B code LLM for exhaustive source-to-sink analysis
A new open-source tool called LVRP (Local Vulnerability Research Pipeline) uses a 14B code-specialized LLM to exhaustively analyze source code for vulnerabilities. It combines code graph and LLM hybrid architecture to enumerate and validate all source-to-sink paths, scaling from small scripts to large codebases like the Linux Kernel and VSCode.
Capn-hook: coding agent tool to avoid grepping the same mystery twice
Capn-hook is a tool for coding agents that records and indexes grep results so that when a developer encounters the same error or mystery again, the agent can recall the previous investigation. It uses LLMs to summarize and store grep outcomes, saving time by preventing repeated searches.
AI-powered tool that auto-generates and syncs code documentation from code changes
A tool that automatically generates and updates documentation by analyzing code changes, ensuring docs always match the actual codebase. It uses LLMs to detect discrepancies and rewrite documentation, solving the problem of outdated docs for developers.
The Patchwork Problem in LLM-Generated Code
A new arXiv paper identifies a structural failure mode in LLM-generated code: individual patches compile and pass tests but break globally due to missing configuration keys, nonexistent imports, or omitted guards. Standard CI toolchains fail to catch these issues, posing a growing risk as LLM coding tools gain adoption.
Enola: engineering intelligence layer for AI coding agents
Enola is an open-source engineering intelligence layer that helps AI coding agents understand existing codebases. It answers questions about change impact, dependency reachability, safe module deletion, refactoring priorities, and architecture drift. The tool uses LLMs to analyze code context and provide insights that reduce mistakes from both humans and AI agents.
