Research paper catalogs software engineering activities being packaged as reusable AI agent skills
A new arXiv paper systematically analyzes which software engineering activities are being encapsulated into reusable agent skills, as emerging skill marketplaces and repositories enable developers to package and share SE expertise. The study addresses a gap in existing research, which has focused on skill acquisition, safety, or benchmarking rather than the specific SE activities being reused.
Entities
Related
Collection of 48 working AI agent examples in Python and TypeScript
A curated repository of 48 functional AI agent implementations covering common patterns like research, code review, SQL, data analysis, and web scraping. Each example is designed to be cloned and run immediately, solving the problem of broken or incomplete agent tutorials for developers building AI systems.
Agent Harness Engineering shared on Hacker News
A post titled 'Agent Harness Engineering' was shared on Hacker News on July 12, 2026, linking to addyosmani.com. The post discusses engineering aspects of agent harnesses, but no further details are provided in the excerpt.
Terry Tao demonstrates coding agents for legacy app modernization
Mathematician Terry Tao published a blog post showing how modern AI coding agents can be used to update and modernize old software applications. The post demonstrates practical techniques for leveraging AI to refactor legacy code, highlighting the potential for AI-assisted software maintenance.
Long-Horizon-Terminal-Bench: New Benchmark Tests AI Agents on Long-Horizon Tasks with Dense Rewards
Researchers introduced Long-Horizon-Terminal-Bench, a benchmark of 46 long-horizon tasks across nine categories, including experiment reproduction and software engineering. Unlike existing benchmarks that evaluate only final outcomes, it uses dense reward signals to measure intermediate progress, providing a more complete picture of agent capability.
temporal-debug-skill project enables AI agents to time-travel debug via git worktree
A new open-source project called temporal-debug-skill gives AI coding agents the ability to debug code at any point in git history. The skill parses fuzzy time references into commit SHAs, creates isolated read-only worktrees, and auto-cleans up, working with any agent like Claude Code or Cursor.
