Gentle-AI is an open-source toolkit that helps engineering teams make the coding agents they already use more consistent, capable and accountable. It works with more than 15 agent runtimes, including Claude Code, Cursor, OpenCode, Codex, Gemini CLI, VS Code Copilot and Pi. Installed as one binary per machine, it configures each agent’s native settings rather than replacing it or locking teams into a new platform.

The stack combines persistent memory, curated skills, adaptable workflows and evidence-based review. Its Engram memory service stores decisions, bug fixes and conventions in a searchable local database, while skills are loaded when relevant. Organic Driven Development is the default: agents explore before editing, match the process to the task, keep changes focused and verify their work. Spec-Driven Development is available when a project benefits from agreed specifications and staged implementation. Additional features include current documentation through Context7, model routing, supported subagents, configurable personas and safeguards for sensitive files.