---
name: Nowledge Mem
slug: nowledge-mem
category: AI Engineering
description: Nowledge Mem provides external memory for Alma, including startup context, broad and focused recall, thread history lookup, and durable memory saves. Use it when a conversation depends on prior decisions, preferences, or stored project facts.
github: "https://github.com/nowledge-co/community/tree/main/nowledge-mem-alma-plugin/skills/nowledge-mem"
language: Python
stars: 166
forks: 35
install: "npx degit https://github.com/nowledge-co/community/tree/main/nowledge-mem-alma-plugin/skills/nowledge-mem ~/.claude/skills/nowledge-mem"
installs_to: ~/.claude/skills/nowledge-mem
source_path: nowledge-mem-alma-plugin/skills/nowledge-mem/SKILL.md
collection_size: 25
category_size: 3670
collection_url: "https://dirskills.com/collections/nowledge-co/community"
added: 2026-09-08T05:34:50.305Z
last_synced: 2026-09-08T05:34:50.305Z
canonical_url: "https://dirskills.com/skills/nowledge-mem"
---

# Nowledge Mem

Nowledge Mem provides external memory for Alma, including startup context, broad and focused recall, thread history lookup, and durable memory saves. Use it when a conversation depends on prior decisions, preferences, or stored project facts.

**Install:**

```bash
npx degit https://github.com/nowledge-co/community/tree/main/nowledge-mem-alma-plugin/skills/nowledge-mem ~/.claude/skills/nowledge-mem
```

## README

# Nowledge Mem

Use Nowledge Mem as Alma's external memory system.

## First Move

At the start of work that may depend on identity, active scope, Rules, recent priorities, or prior decisions, call `nowledge_mem_context_bundle`.

Use `nowledge_mem_working_memory` only when you need the lighter daily briefing or when Context Bundle is unavailable.

## Retrieval

- Use `nowledge_mem_query` for broad recall across memories, with thread fallback.
- Use `nowledge_mem_search` for focused memory retrieval.
- Use `nowledge_mem_thread_search` when the user is asking about conversation history.
- When a memory includes `sourceThreadId`, use `nowledge_mem_thread_show` to inspect the original conversation before making strong claims.

Start with a narrow query. Fetch more pages only when the current result is not enough.

## Saving

Use `nowledge_mem_store` when the conversation produces durable knowledge:

- decisions
- user preferences
- debugging conclusions
- workflow agreements
- plans that should survive the current chat
- facts about the user's projects or work

Choose `unit_type` deliberately: `decision`, `learning`, `preference`, `fact`, `plan`, `procedure`, `context`, or `event`.

Do not save pleasantries or filler. If the user explicitly asks you to remember something, save it.

## Corrections And Cleanup

- Use `nowledge_mem_update` when a remembered item is wrong but should remain.
- Use `nowledge_mem_delete` only when the user clearly wants a memory removed.
- Use thread delete tools only on explicit user intent.

## If Tools Are Hidden

If Alma does not expose plugin tools in the current thread, use Bash with `nmem` when available:

```bash
nmem --json context read --source-app alma
nmem --json m search "<query>" -n 5
nmem --json t search "<query>" -n 5 --source alma
nmem --json t show <thread_id> -n 30 --offset 0 --content-limit 1200
nmem --json m add "<content>" -t "<title>" --unit-type decision
```

If neither plugin tools nor Bash are available, state the blocker and ask for one concrete enablement step.

## Response Contract

When memory affected the answer, include a short source line:

- `Source: nowledge_mem_query`
- `Source: nowledge_mem_search + nowledge_mem_show`
- `Source: nowledge_mem_thread_search + nowledge_mem_thread_show`
- `Source: nmem CLI`
- `Source: injected recall context`