Prismer Setup

Step 1: Check existing config

cat ~/.prismer/config.toml 2>/dev/null | grep api_key | head -1 || echo "No config found"

If already configured and user doesn't want to reconfigure, stop here.

Step 2: Run auto-setup

IMPORTANT: This command opens the browser and waits for the user to sign in. Set Bash timeout to 300000 (5 minutes):

node "${CLAUDE_PLUGIN_ROOT}/scripts/setup.mjs" --force

The script:

1. Starts a local callback server
2. Opens the browser to prismer.cloud/setup
3. User signs in or registers → key auto-created → redirected back to localhost
4. Key saved to ~/.prismer/config.toml

Wait for it to complete. It will print "API key saved" when done.

Step 3: Mark setup complete

CACHE="${CLAUDE_PLUGIN_DATA:-${CLAUDE_PLUGIN_ROOT}/.cache}"
mkdir -p "$CACHE"
echo "$(date +%s)" > "$CACHE/.setup-done"

Tell the user: "Setup complete! Run /reload-plugins to activate hooks."

Step 4: MCP Tools (optional)

Ask the user: "Do you also want MCP tools? These let Claude proactively call evolution/memory/skill APIs (29 tools). Without MCP, hooks still work (passive context injection). With MCP, Claude can actively create genes, write memory, search skills."

If yes:

claude mcp add prismer -- npx -y @prismer/mcp-server@1.8.0

If no: skip. Plugin works fine without MCP — hooks handle sync/journal/stuck-detection independently.