Technical-preview local graph memory for DSH and Codex: recalls owner-global and current-project evidence before each turn, writes USER and ASSISTANT records separately, and preserves project, session, actor, and source provenance for a visual atlas and cited knowledge pages.
Install
# from GitHub (first run asks for allowBuilds approval — follow the hint, retry)
dsh plugin --profile web add github:reshuibuduo/TMCRA-Agent-Memory#path:/integrations/deepseek-harness-local
Any plugin you install runs third-party code with your own permissions — it can read your files, use your credentials, and reach the network, and tool approvals don’t sandbox it. GitHub-sourced plugins also run build scripts at install time. Only install sources you trust, and pin a commit (github:owner/repo#sha).
README
This DSH plugin connects the native agent/pre-step and session/event lifecycle seams to a TMCRA instance running on the same computer.
It requires the owner-local runtime and the shared integration config created by:
powershell -ExecutionPolicy Bypass -File .\scripts\install-local.ps1
node .\integrations\local-agent-hooks\scripts\configure.mjs --runtime-config .\.tmcra\config\runtime\local-runtime.json
Install and test the local package, then add it to the selected Harness profile:
powershell -ExecutionPolicy Bypass -File .\scripts\install-deepseek-harness-local.ps1 `
-PackageDirectory D:\tmcra-packages
Shell:
TMCRA_DSH_PACKAGE_DIRECTORY="$HOME/tmcra-packages" \
bash scripts/install-deepseek-harness-local.sh
The installer runs npm ci, type checking, the Harness lifecycle test, the build, npm pack, and dsh plugin --profile web add. The plugin reads ~/.tmcra/local-integration.json by default. configPath can point to another owner-local integration config. DeepSeek Harness preview releases can mis-handle tarball paths containing spaces or non-ASCII characters, so the packaging directory must be a short ASCII-only path.
For every admitted human turn, the plugin:
- resolves a project identity shared with the other local TMCRA adapters;
- recalls user-global and current-project evidence before the first model step;
- injects that evidence as an explicit plugin-sourced, untrusted user message;
- stores the human prompt with
userprovenance; - stores the visible completed answer with the exact Harness agent/subagent identity;
- queues failed writes locally and retries them on the next turn.
No TMCRA account, subscription, device authorization, hosted scope, or production API endpoint is part of this package.
The lifecycle and package contents have been accepted against @deepseek-ai/dsh-agent-loop 0.1.0-rc.6. The final dsh plugin add command still requires a locally installed DSH CLI; the installer fails if it is missing.
Links
More in this category
volcengine/OpenViking#examples/dsh-memory-plugin★ 28995
OpenViking memory and context bundle for DeepSeek Harness: pre-step auto-recall and profile injection, session capture, `viking://` URI guarding, and recall/write memory tools backed by an OpenViking server.
vectorize-io/hindsight#coding-agents★ 20139
Hindsight, agent memory that learns: long-term project memory with auto recall and retain, knowledge pages, deep reflection, and per-repo memory banks.
Ikalus1988/MisakaNet★ 405
Failure-recovery memory: search and record failure-recovery lessons from real engineering sessions, with BM25 + semantic RAG retrieval and a lessons knowledge base.
text2future/flowix#dsh-flowix-memory★ 316
Registers the local flowix-cli MCP server so the agent can search, read, create, and edit Flowix memos and mind-map artifacts.
dsh-engramory★ 155
The Engramory curated-memory discipline as an installable plugin ([npm: dsh-engramory](https://www.npmjs.com/package/dsh-engramory)): a deterministic 200-line / 25 KB cap on the `MEMORY.md` index via `ctx.tools.guard()` — growth denied, a shrinking rewrite always passes — plus the protocol registered as a runtime skill. The store is plain markdown, one file per fact, shared with Claude Code, Codex, Kiro, and OpenClaw.
omdsh-dev/dsh-mnemon★ 92
Cross-agent, local-first persistent memory plugin for DeepSeek Harness (DSH), powered by Mnemon. It shares long-term memory across Mnemon-enabled agents and adds runtime memory, searchable project documents, semantic recall, knowledge graph, and a Sidebar UI.