Mem setup

Skill akunzai/agent-skills/skills/mem-setup

My personal agent skills for AI coding assistants

Install

npx -y skills add akunzai/agent-skills --skill mem-setup

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Wire each installed coding agent's native global memory file to one canonical ~/.agents/AGENTS.md, so Claude Code, Codex, OpenCode, Antigravity/Gemini, and pi all read the same durable cross-agent memory. Use to set up or refresh global memory bridging across agents.

SKILL.md

3.7 KB, 872 tokens by cl100k_base, as published. Nobody here has run it

mem-setup — Cross-Agent Global Memory Bridge

Points every installed agent's global memory at a single canonical core file so durable instructions are shared across agents. Ships the wiring only — you supply the content of ~/.agents/AGENTS.md (and any augmentation modules).

Canonical layout (user-owned)

~/.agents/AGENTS.md — the cross-agent core, the single source of truth.
Optional augmentation modules — extra rules inlined for agents with weak hook/skill support. Pass their paths via MEM_SETUP_AUGMENT (colon-separated). The repo ships none.

Per-agent bridge

Agent	Native global path	Tier	Method
Claude Code	`~/.claude/CLAUDE.md`	high (core)	`@import`
Codex	`~/.codex/AGENTS.md`	high (core)	symlink (copy on Windows)
pi	`~/.pi/agent/AGENTS.md`	high (core)	symlink (copy on Windows)
Antigravity / Gemini	`~/.gemini/GEMINI.md`	low (core + augment)	`@import`
OpenCode	`~/.config/opencode/opencode.json`	low (core + augment)	`instructions[]`

High-tier agents get dynamic behaviors from plugins, so they read the core only. Low-tier agents have weak hook support, so augmentation modules are composed into their static prompt via import / config — never by duplicating a file.

Gemini's @import method degrades to a plain symlink when there's nothing to compose (no MEM_SETUP_AUGMENT) and the target is missing or already a symlink — same effect, but skips depending on the target agent's own import resolution. A configured augment module always keeps the append-import behavior (a symlink can't reference two files). An existing real GEMINI.md keeps it too, unless its bytes exactly match this script's own prior stub output (no augment) — that case is self-healed into a symlink (after a backup), since it holds no manual content to preserve. This only applies to the low tier — Claude Code's own import bridge always stays an @import stub, never a symlink.

OpenCode's config method edits instructions[] in JSON: it prefers jq (the tool this repo's mise.toml pins), falls back to python3 if jq is absent, and otherwise prints the paths to add manually. Neither is a hard requirement.

Workflow (always dry-run first)

Ensure ~/.agents/AGENTS.md exists (your durable cross-agent instructions).
Run the read-only plan and show it to the user: bash skills/mem-setup/scripts/mem-setup-bridge.sh plan
After explicit user confirmation, apply: bash skills/mem-setup/scripts/mem-setup-bridge.sh apply

Override the core path with MEM_SETUP_CANONICAL and augmentation modules with MEM_SETUP_AUGMENT (colon-separated absolute paths).

Safety

Only acts on installed agents (those whose config directory exists).
Never write through a symlink — an existing symlink is removed first, so the canonical core is never clobbered.
Real files are backed up to *.bak-<timestamp> before any change.
Idempotent — re-running makes no changes once bridged.
Windows (Git Bash): import/config methods need no symlink; for Codex/pi the script falls back to a copy and warns that you must re-run to refresh.

Migration from `memory-autoload`

This skill replaces the retired memory-autoload plugin and its ~/.agents/MEMORY.md. If that plugin is still installed, remove it; if ~/.agents/MEMORY.md holds content, merge it into ~/.agents/AGENTS.md first.