Wavestone AI Lab
An AI research laboratory under Wavestone that focuses on source-code-level analysis and architectural research of enterprise-grade AI systems.
Timeline (last 90 days)
Wavestone AI Lab published an 83-page research report conducting a source-code-level teardown of 11 production-grade coding Agents.
The report covers 11 systems in total—Claude Code, Codex CLI, Gemini CLI, Mistral Vibe, OpenHands, Aider, Mini-SWE-Agent, Hermes, Pi, OpenCode, and OpenClaw—plus the meta-framework control group Omnigent.
The study involves approximately 4 million lines of Python, TypeScript, and Rust code, combined with 90 days of longitudinal diff analysis.
The evaluation numbers appearing in the paper come from the system vendors themselves and cannot be rerun locally; the benchmark scores were not independently verified.
Checklist-type assertions (number of tools, feature cells, version pinning) decay on a weekly basis, whereas structural assertions (loop taxonomy, subsystem anatomy, missing items) have so far proven durable.
All Facts (5)
The report covers 11 systems in total—Claude Code, Codex CLI, Gemini CLI, Mistral Vibe, OpenHands, Aider, Mini-SWE-Agent, Hermes, Pi, OpenCode, and OpenClaw—plus the meta-framework control group Omnigent.
50%UnverifiedThe study involves approximately 4 million lines of Python, TypeScript, and Rust code, combined with 90 days of longitudinal diff analysis.
50%UnverifiedThe evaluation numbers appearing in the paper come from the system vendors themselves and cannot be rerun locally; the benchmark scores were not independently verified.
50%UnverifiedChecklist-type assertions (number of tools, feature cells, version pinning) decay on a weekly basis, whereas structural assertions (loop taxonomy, subsystem anatomy, missing items) have so far proven durable.
50%UnverifiedWavestone AI Lab published an 83-page research report conducting a source-code-level teardown of 11 production-grade coding Agents.
50%