Recommended Setup
Proma uses three parts: MCP tools for intelligent retrieval, lifecycle hooks for automatic thread capture, and a managed Nowledge Mem block in the workspace CLAUDE.md for startup context.
Command
rm -rf /tmp/nowledge-community
git clone https://github.com/nowledge-co/community.git /tmp/nowledge-community
mkdir -p ~/.proma/scripts ~/.proma/agent-workspaces/default/skills
cp /tmp/nowledge-community/nowledge-mem-proma-plugin/hooks/save-to-nmem.py ~/.proma/scripts/
cp /tmp/nowledge-community/nowledge-mem-proma-plugin/hooks/read-working-memory.py ~/.proma/scripts/
chmod +x ~/.proma/scripts/save-to-nmem.py ~/.proma/scripts/read-working-memory.py
cp -R /tmp/nowledge-community/nowledge-mem-proma-plugin/skills/{read-working-memory,search-memory,distill-memory,save-thread,status} ~/.proma/agent-workspaces/default/skills/Then add the Nowledge Mem MCP server to ~/.proma/agent-workspaces/default/mcp.json and merge the plugin hooks into ~/.proma/sdk-config/.claude/settings.json. Use the detailed guide for the exact JSON blocks.
What Happens Automatically
- SessionStart refreshes the Nowledge Mem block in Proma's workspace
CLAUDE.md - UserPromptSubmit and Stop hooks parse Proma SDK JSONL sessions and append messages into Mem threads
- Repeated hook runs deduplicate by Proma message UUID
- Stop asyncRewake can push a compact Working Memory reminder after assistant turns
- MCP tools let Proma search memories, save durable knowledge, search threads, and check status
- Skills teach the agent when to read Context Bundle or Working Memory, search, distill, save-thread, and check status
MCP recall and distillation are guided by the agent. Thread capture and startup context are lifecycle hooks, so they do not depend on the model choosing a tool.
Verify
Restart Proma, ask What was I working on?, then send a message with a unique phrase. After Proma replies:
Command
nmem t search "that phrase" --source promaThe Proma thread should appear in Mem.
