nowledge-co/community

The Nowledge Mem community-integrations repo ships a native DeepSeek Harness bundle: Context Bundle injection, prompt-time recall, Mem MCP tools and turn-end transcript capture.

This repo collects community integrations for Nowledge Mem (mem.nowledge.co). The DeepSeek Harness entry is described in the README's own words as a 'Community DSH bundle with native Context Bundle injection, prompt-time recall, Mem MCP tools, and session/event turn-end transcript capture through nmem'. The README keeps one shared language across hosts so a fresh user can tell 'tools available' apart from 'autonomous memory behaviour', and points to shared/behavioral-guidance.md for how the bundle decides when to read, search, save and route memory.

Memory ★ 185
View on GitHub ↗

Install

dsh plugin --profile web add github:nowledge-co/nowledge-mem-deepseek-harness

Copied verbatim from the README's 'DeepSeek Harness Plugin' integration-table row. The README's registry file (integrations.json) is the canonical source of truth for every integration; the DeepSeek Harness entry points at the standalone subdirectory nowledge-mem-deepseek-harness, which the README says is tagged dsh-plugin for ecosystem discovery.

Compatibility

The README's integration registry documents an 'autonomy contract' with three levels — automatic (the host or plugin enforces behaviour through hooks or lifecycle wiring), guided (rules and skills teach it but the model decides) and manual (only on explicit request) — so the DSH bundle's behaviour is described rather than promised. Multi-agent orchestrators set NMEM_AGENT_ID=<agent-slug> per child process and only add NMEM_SPACE=<space> to override an AI Identity's default space; NMEM_HOST_AGENT_ID is for advanced external aliases.

Details

FAQ

Does the bundle capture sessions automatically?
The README describes the DeepSeek Harness bundle as capturing turns through a session/event turn-end hook via nmem; across all integrations it labels each behaviour as automatic, guided or manual so the difference between tools-available and automatic behaviour stays honest.
How do multi-agent setups keep identities separate?
The README says to set NMEM_AGENT_ID=<agent-slug> in each child process, and to add NMEM_SPACE=<space> only when a run should override the AI Identity's default space.

Alternatives

diqierjia/StrataGate-AgentMemory · Co-Engram/Co-Engram · ICCuse/dsh-file-memory

More plugins in Memory

Browse more in Memory

Guides for Memory plugins