Skip to content
This repository was archived by the owner on Aug 18, 2026. It is now read-only.

Repository files navigation

ren

A five-petaled lotus rooted in memory, with a continuous flow connecting bloom and roots

記憶に根を張り、開発を連ね、智を錬る。

ren is a foundation for continuous development with coding agents. It currently provides deterministic Rhai workflows and durable local memory, and is intended to grow across other parts of the development flow.

The five petals

At the center of ren is the 5-Step Engineering Process. Its five steps form the petals of a lotus and must be followed in order:

Petal Step Practice
1 Make your requirements less dumb Question every requirement and make it earn its place.
2 Delete the part or process Remove anything that does not need to exist.
3 Simplify / Optimise Improve only what survives deletion.
4 Accelerate cycle times Shorten the path from action to feedback.
5 Automate Automate the cycle only after the earlier steps have shaped it.

The process does not end at the fifth petal. What the workflow learns returns to memory, and the next cycle begins with better-grounded requirements.

The name

The name ren carries three connected ideas:

  • 蓮 (lotus) — in the Five Phases, water represents wisdom; the lotus roots in memory and lets the five-petaled process bloom from muddy information.
  • 連 (connection) — linking people, agents, context, and steps into a continuous development flow.
  • 錬 (refinement) — returning workflow experience to memory so that knowledge can be refined over time.

ren-workflow

Discover the available workflows and their arguments before running one:

ren workflow list
ren workflow show <name>
ren workflow schema <name>
ren workflow run <name-or-path> --args '<json>'

Run ren workflow --help for the complete command reference.

The ren init command installs the embedded skills into every supported coding agent — Claude, Cursor, Codex, Grok, OpenCode, and pi — or a single one with ren init --agent <agent> (user scope by default, --project for the current repository). pi reads user-global skills from ~/.pi/agent/skills and project skills from .pi/skills.

The bundled implement workflow connects implementation work back to project memory. It searches and inspects relevant prior notes, captures bounded fleeting notes after implementation and fix passes, checks the change for contradictions with remembered decisions during independent review, and promotes verified knowledge through an inspect-then-explicitly-apply proposal. Its final report is ready to adapt into a PR description and includes the memory operations used. Projects without a registered memory vault still run; the report records that memory work was skipped. Task text is limited to 500 characters so the maximum-effort, eight-round execution plan remains bounded.

ren-memory

Capture and retrieve durable project knowledge through the local ren-memory vault:

ren memory --help
ren memory search <query>
ren memory list
ren memory show <note-id>

The hyphenated ren-memory name refers to the skill and component; ren memory is its CLI command group.

Set up the Codex hook

From the project directory whose memory vault you want to register, run these commands in order:

ren memory init --user
ren memory index --rebuild
ren memory hook install --agent codex --user

The hook is installed in the configuration for the current CODEX_HOME. When CODEX_HOME is set, the target is $CODEX_HOME/config.toml; otherwise it is $HOME/.codex/config.toml. Run the hook installation command once for each CODEX_HOME you use, because each one has a separate Codex configuration.

About

A foundation for continuous development with coding agents, featuring deterministic Rhai workflows and durable local memory.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Used by

Contributors

Languages