What problem it solves

AI coding agents lose context between sessions, so the same decisions get re-explained and nothing about the project is remembered. They also lack a shared workflow and evidence, leaving the human to read every line to confirm what was done. This project adds memory, a workflow, and evidence to agents you already have.

Key features

• Persistent project memory with Engram — Engram lets the agent write down what it learns as it goes and consult that memory before asking the user, so context accumulates across sessions instead of resetting.

  Source: README, docs/engram.md

• Proportional workflow with Organic-Driven Development — ODD keeps understood, low-risk changes lightweight and gives substantial authorized work one recoverable feature document. Tasks are sized by understanding, risk, and resumability rather than file count.

  Source: README, docs/usage.md

• Deterministic transitions from a dedicated binary — The gentle-ai binary reads change state from files on disk and returns the only valid next transition, so different agents, models and sessions converge on the same answer.

  Source: README, cmd

• Bounded review of finished work — Receipt-Driven Development freezes the candidate before anything reads it and picks a review depth from that frozen candidate, keeping the result informational and leaving commit, push and release with the user.

  Source: README, docs/review-integration.md

• Configures the agent you already use — The installer configures 17 agents, including Pi, OpenCode, Claude Code, Codex and Cursor, using each agent's native capabilities, and never installs an AI agent for you. Configs are snapshotted before every write.

  Source: README, docs/agents.md

Quick start

Copied verbatim from the project's README.

gentle-ai          # pick your agents, components and persona
gentle-ai doctor   # verify — read-only, changes nothing

Who it may suit

Inference: this may suit developers and teams already using AI coding agents who want persistent memory, a shared workflow, and review evidence without switching tools or committing to one vendor.

Topics to discuss

• How should persistent agent memory be scoped and revoked across projects and clones?

• Which routing triggers for delegation feel right for your repository, and where does inline work break down?

• How should bounded review depth be calibrated so review stays informative without becoming a bottleneck?

Written by ReadmeX AI from the project's README, source and public GitHub data; it may contain mistakes.

Read the original README https://github.com/Gentleman-Programming/gentle-ai#readme

ReadmeX opened this community for a trending project; it is not run by the project. Maintainers can claim it from the community page.