Manual Copy
Last commit on March 20, 2026·Created on March 19, 2026
GODGOD126/self-improving-for-codex
“Wires five memory files into a loop Codex can follow.”
Combined rank
#517
across all skills
In Agent Orchestration
#30
category rank
Stars
128
+0.8% in last 7d
Forks
17
0.0% in last 7d
Watchers
1
0.0% in last 7d
Traction scoreGitHub stars can be faked, so popularity alone can be misleading. Traction Score looks for broader signs of real attention, adoption, and active maintenance.
Description
This skill gives Codex a self-improving memory loop. It creates five persistent files—PROFILE, ACTIVE, LEARNINGS, ERRORS, and FEATURE_REQUESTS—in a global directory, then wires them through AGENTS.md so Codex reads them before each task. An optional nightly automation refines these files automatically. It serves Codex users who want durable cross-session memory and automation engineers who want hands-off memory maintenance.
WHO IT'S FOR
Codex users
set up persistent memory loop
AI workflow engineers
design memory architecture for AI assistants
AI Tooling Leads
configure shared memory rules for team
Automation engineers
implement nightly memory refinement automation
30 Days of GitHub Stars
+2 stars↑ 1.6% monthly change
Jul 7Jul 14Jul 21Jul 28Aug 5
Compatibility & Install1 Skill · 2 Compatible agents · 3 categories · 1 install path
Compatible agents
❯Codex❯OpenClaw
Categories
Agent OrchestrationAutomationKnowledge
Install via
Repository instructions
Also in this category4 of 32