add-memory

Add Memory

ℹ️ First, set expectations: before anything else, print one short line with this skill's version and its most recent change — the top entry of metadata.changelog above — e.g. add-memory vX.Y — recent: <summary>. Then proceed.

Add a memory system to an existing agent. Memory skills are COPIED into the agent's .claude/skills/ directory — the agent becomes self-contained with no plugin dependency.

Memory Types

Type	Description	Skills Installed
file-index	File system awareness — creates and maintains a searchable file index	setup-index, refresh-index, search-files
brain	Zettelkasten-style knowledge graph — atomic notes with connections	setup-brain, create-note, search-brain, find-connections
json-state	Structured JSON state with jq-based updates and action logging	setup-memory, load-memory, update-memory, memory-jq
workspace	Multi-session project tracking — organized folders, sessions, archives	setup-projects, create-project, create-session, archive-project

Process

Step 1: Determine Agent Directory

Check if we're in an agent directory (has CLAUDE.md) or if user specified a path.

# Check current directory
if [ -f "CLAUDE.md" ]; then
  AGENT_DIR="$(pwd)"
else
  # Ask user for agent path
fi

If no CLAUDE.md found, ask:

• Question: "What agent should I add memory to?"
• Hint: "Enter the path to the agent directory (should contain CLAUDE.md)"

Step 2: Check Argument or Ask Memory Type

If user provided an argument (e.g., /agent-dev:add-memory brain), use that.

Otherwise, use AskUserQuestion:

• Question: "What kind of memory does this agent need?"

• Header: "Add Memory"

• Options:

  1. File awareness (file-index) — "I need the agent aware of workspace files"

     • Creates a searchable index of files in the workspace
     • Good for: agents that need to find and reference files

  2. Knowledge graph (brain) — "I need connected notes, evolving knowledge"

     • Zettelkasten-style atomic notes with connections
     • Good for: research agents, knowledge management, idea linking