graph-memory DSH plugin

Graph Memory is a cross-session memory plugin for DeepSeek Harness. It extracts typed nodes and edges from conversations into a local SQLite store, then retrieves a small relevant subgraph during prompt assembly, so recall survives restarts and stays traceable back to its source session.

developer-tools skills-workflows

How to install the graph-memory DSH plugin

npx @deepseek-ai/dsh plugin --profile web add /absolute/path/to/graph-memory-1.6.0-beta.1.tgz

Copying does not run this command. Review the repository and version before use.

Data source

Snapshot date: Aug 16, 2026

discovered

What the graph-memory DSH plugin can do

Extracts TASK, SKILL and EVENT nodes from conversations and links them with typed edges such as USED_SKILL, SOLVED_BY, REQUIRES, PATCHES and CONFLICTS_WITH.
Recalls memory automatically during DSH prompt assembly, so no explicit search tool call is needed to bring past context back.
Ranks and bounds retrieval with community detection, PageRank, personalized PageRank and bounded graph traversal, keeping only a local subgraph in context.
Falls back to SQLite FTS5 lexical search when no embedding provider is configured, so retrieval keeps working without an API key.
Exposes gm_status, gm_search, gm_record and gm_stats as DSH tools for inspecting and writing memory by hand.
Hooks into the DSH/Cordis plugin lifecycle and disposes its database, cache and listeners with the plugin fiber instead of patching harness core.

Where the graph-memory DSH plugin fits

Carrying a debugging session's findings into a later session without pasting the earlier transcript back in.
Building a reusable store of validated fixes so a recurring error resolves from recorded experience rather than a fresh investigation.
Keeping long-running project context available after DSH restarts, where compaction would otherwise drop it.
Auditing why a piece of context was recalled, by following the graph edges and source session back to the original message.

Who the graph-memory DSH plugin is for

DSH users who work across many sessions on the same codebase and keep re-explaining the same background.
Developers who want retrieval that is inspectable, rather than an opaque vector blob.
Teams running DSH locally who need memory to stay on their own machine by default.

graph-memory DSH plugin limitations

The current release is a beta that is not published to npm, so installation requires cloning the repository, running the build and packing a tarball yourself.
The install command points at an absolute path to that tarball, which means the command shown here cannot be copied and run unchanged.
Requires Node.js 22.19 or newer; older runtimes are not supported.
Vector retrieval needs an external embedding provider and API key. Without one the plugin still runs, but falls back to lexical search only.
Graph Memory Pro is documented as OpenClaw-based and is not installable into DSH today; only the community edition is a DSH plugin.

From the repository README

Quoted from the adoresever/graph-memory README. Copyright remains with the original authors.

Compaction answers "how much of this conversation still fits?" Graph Memory answers "which past knowledge is worth recalling now?" Reusable conversation knowledge becomes typed nodes: `TASK` for goals, execution and outcomes; `SKILL` for validated reusable methods; `EVENT` for errors, fixes, decisions, changes and facts. Typed edges such as `USED_SKILL`, `SOLVED_BY`, `REQUIRES`, `PATCHES` and `CONFLICTS_WITH` preserve relationships. A new question retrieves a relevant local subgraph instead of replaying the complete history. Native host integration: loaded by the DSH/Cordis plugin lifecycle, not simulated through an MCP side channel. Disposes database, cache and event listeners with its plugin fiber. Does not fork or modify DeepSeek Harness core.

Read the full READMERepository license: MIT

graph-memory DSH plugin questions

What is the graph-memory DSH plugin?

graph-memory is a project included in the current DSH plugin snapshot. This page organizes repository metadata and reviewed notes, but you should still inspect the upstream repository before running it.

How do I install the graph-memory DSH plugin?

The page shows the install command captured from the source record. Copying does not run it, so confirm the package, profile, and version against upstream documentation first.

Which DSH profile does graph-memory use?

The current snapshot does not confirm a profile for graph-memory. Check its manifest and README so the plugin is not added to the wrong DSH profile.

Is graph-memory compatible with the current DSH release?

The source data does not include a verified compatibility matrix. Check recent commits, releases, and DSH breaking changes, then test in a profile you can roll back.

What should I do if graph-memory fails to load?

Check the install command, selected profile, dependency logs, and Cordis startup output first. Keep the exact error string, then compare it with this site's fix pages and upstream issues.

What are the closest alternatives to graph-memory?

Related plugins on this page come from the current snapshot and are only candidates. Compare capabilities, maintenance activity, installation method, and limitations before choosing one.