init

Configures per-project memory storage mode and writes the memory plugin config.

20|2|Updated May 16, 2026
One-click install
npx skills add https://github.com/reddb-io/red-skills --skill init-reddb-io
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: init
Source: https://github.com/reddb-io/red-skills/tree/main/plugins/memory/skills/core/init
Command: npx skills add https://github.com/reddb-io/red-skills --skill init-reddb-io

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Setting up persistent agent memory for a project requires choosing a storage mode, writing a valid config block, and deciding which hooks should fire — doing this by hand risks invalid config and misconfigured hooks. This Skill runs a guided one-time wizard that writes the per-project memory config correctly. ## Core Features & Use Cases - Storage mode selection: Choose between markdown-only (plain notes under .red/memory/notes/, no engine) and graph mode (governed operational memory over a per-project RedDB store at .red/memory/graph.rdb). - Automatic runtime bootstrap: The bundled bootstrap.mjs script downloads the memory CLI and native red engine on first use with checksum verification — no build step required. - Optional auto-firing hooks: In graph mode, optionally enable SessionStart recall, PostToolUse re-index, Stop extract, and PreCompact flush hooks, which stay dormant until the config flag is on. - Use Case: After installing the memory plugin, run the wizard to pick graph mode with hooks, then immediately use /memory:store and /memory:recall against the configured store. ## Quick Start Ask the assistant to initialize memory for this repository and walk you through choosing a storage mode.

Frequently Asked Questions about init

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I set up memory for a project with the memory plugin?▼

Run the init wizard from the repository root, which asks which storage mode to use and writes the `plugins.memory` block to `.red/config.yaml`. Pass `--mode markdown-only` or `--mode graph` to the bootstrap script, and add `--hooks` only in graph mode if you want auto-firing hooks.

What is the difference between markdown-only and graph memory modes?▼

Markdown-only stores plain searchable notes under `.red/memory/notes/` with no engine, hooks, or MCP server. Graph mode stores governed operational memory as nodes and edges in a per-project RedDB store at `.red/memory/graph.rdb`, with dedupe, provenance, and optional hooks.

Does the memory plugin require installing RedDB separately?▼

No separate installation is needed. The bundled bootstrap script downloads the memory CLI and native `red` engine into a cache directory on first use and verifies checksums. Markdown-only mode does not require RedDB at all.

How do I enable auto-firing memory hooks?▼

Pass `--hooks` during init in graph mode only after the user confirms. Hooks are wired in the plugin manifest but stay dormant until the per-project config flag is on. On Codex, also set `[features].plugin_hooks = true` since hooks are off by default there.

Should the RedDB graph store be committed to version control?▼

No. The `.red/memory/graph.rdb` store is per-project local state and should not be committed, similar to `node_modules/`. The config in `.red/config.yaml` is written through the CLI to keep the schema valid.