dsh-semantic-memory
Persistent semantic long-term memory plugin for DeepSeek Harness.
Install
$ pnpm dsh plugin --profile web add dsh-plugin-semantic-memoryPlugin Overview & Capabilities
AI-assisted organization based on the public repository snapshot. The content must be grounded in source evidence and does not replace compatibility or security verification.
Key Capabilities
- Persists facts, decisions, preferences, and notes across sessions.
- Performs embedding-based semantic retrieval using local ONNX embeddings or an OpenAI-compatible embeddings API.
- Provides memory_write, memory_search, memory_forget, and memory_stats model-facing tools.
- Injects recalled or fallback memories into the system prompt.
- Tags memories with workspace context and supports workspace-scoped recall.
- Applies memory importance, access refreshing, and time-based strength decay.
- Can summarize recent conversation messages into automatically tagged memory entries.
Useful For
- Remember user preferences and durable project decisions between DeepSeek Harness sessions.
- Retrieve relevant prior context by meaning rather than literal full-text search.
- Maintain workspace-specific notes while allowing optional cross-workspace recall.
- Let an agent automatically preserve salient conversation facts and decisions.
Who It Fits
- DeepSeek Harness users who need persistent personal or project memory.
- Agent workflow builders using Cordis-based DSH profiles.
- Users who want local embedding retrieval or OpenAI-compatible embedding endpoints.
Documented Limitations
- The first local embedding downloads an approximately 100 MB model.
- Recall is asynchronous and may arrive one turn late during cold starts or slow API responses.
- A new process may have empty prompt injection before the memory store is lazily loaded.
- Search is brute-force O(n) cosine similarity and is described as suitable for personal-scale stores rather than millions of entries.
- Auto-summarization requires llm and agentDefaultModel services and is silent on failure.
- For file: dependencies, local patch or build changes require reinstalling the profile package and restarting.
DSH Compatibility
Version-specific runtime evidence collected by DSH Plugin. A missing result means we have not tested that combination yet.
Security Signals
Objective signals discovered from package metadata and source inspection. These are not a guarantee that a plugin is safe.
package.json declares 1 runtime, 5 development, 3 peer, and 0 optional dependencies.
package.json declares DSH bundle metadata.
GitHub reports the repository license as Apache-2.0.
A root package.json was captured and can be inspected by the registry.
Public GitHub source metadata is available for this registry snapshot.
Source & Registry Notes
Public provenance, Registry classification, and the latest source check for this entry, kept separate from runtime verification.
- Source repository
- chenkezhen480/dsh-semantic-memory
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin
- Source checked
- 0ab4575 · 2026-09-15
This project is independently indexed from public source information. DSH Plugin is not affiliated with DeepSeek or the plugin author. Always check the author repository before installation.
Repository Activity
- GitHub stars
- 0GitHub stars
- Forks
- 0
- Open issues
- 0
- Last commit
- 2026-08-17
- Last release
- No release detected
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
Bundle of 88 open-source Agent Skills for research, business intelligence, marketing, and SEO in DeepSeek Harness.
View plugin →Skill and regex/pattern linter for detecting and humanizing AI-generated Russian text in DeepSeek Harness.
View plugin →A DeepSeek Harness Cordis skill plugin that packages 85 reverse engineering, authorized pentesting, security research, and CTF SKILL.md resources.
View plugin →A standalone, local persistent-memory system for AI agents, with DeepSeek Harness support provided through the separate dsh-mnemon plugin.
View plugin →