ai-agents-research
Field research and feature analysis for AI coding agents — sandboxing, orchestration, plugins, community tooling, SDLC patterns.
A continuously-updated research catalog: make informed adopt / defer / skip decisions about coding agents and their ecosystems before building production systems on them. Tracks Claude Code, JetBrains Air, DeerFlow, Goose, Codex, Devin and others — plus the surrounding plugins, observability, SDLC patterns, and cross-repo learnings.
Each analysis follows the same four-section structure:
What it is → How it works →
Adoption decision → Action items. Currency is
maintained by four cron-driven monitors.
Contents
cc-native
Anthropic-native features: agents/skills, CI/sandboxing, context/memory, configuration, plugins, model-internals, MCP.
non-cc
Non-CC agents, orchestrators, and infrastructure: JetBrains Air, DeerFlow, Goose, Rowboat, and more.
cc-community
Community skills, plugins, tooling, and domain-specific CLAUDE.md patterns.
sdlc-lcm
SDLC / lifecycle-management specs, agentic SDLC patterns, OSS ALM landscape.
learnings
Cross-repo compound learnings — recurring patterns from live development across the qte77 ecosystem.
research
Auto-generated cumulative index of agentic-AI papers from the rxiv eval pipeline.