dsh-local-vector-memory
Fully local SQLite vector memory plugin for DeepSeek Harness with automated recall injection and deduplication.
Install
$ dsh plugin --profile web add dsh-local-vector-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
- Stores and searches long-term memories locally using SQLite and local embedding endpoints
- Automatically intercepts memory trigger words in user messages for capture
- Injects top-K relevant memories into agent context prior to LLM inference with LRU deduplication
- Detects potential conflicts and near-duplicate memories (similarity threshold check)
- Supports soft-delete recycle bin with restoration capabilities and permanent purging
- Provides online SQLite database backups via VACUUM INTO
- Falls back to keyword matching if the embedding service is unreachable
- Offers 10 dedicated DSH tools for memory management, extraction, and reindexing
Useful For
- Persisting user preferences, facts, and conversation context across DSH sessions without cloud dependency
- Enhancing DeepSeek Harness agents with automated local RAG and memory retrieval
- Managing and editing past conversational memories with conflict detection and recycle bin safety
Who It Fits
- DeepSeek Harness users seeking offline, privacy-first persistent memory
- Developers building personalized local AI workflows with custom embedding setups
Documented Limitations
- Requires Node.js >= 22.5.0 for built-in node:sqlite support
- Requires an external OpenAI-compatible local embedding service (e.g., Qwen3-Embedding on llama-server)
- LLM-based automatic extraction (autoExtract) is disabled by default due to token and reasoning chain constraints on long texts
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 0 runtime, 0 development, 2 peer, and 0 optional dependencies.
package.json declares DSH bundle metadata.
GitHub reports the repository license as MIT.
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
- liangxiaobing520/dsh-local-vector-memory
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin
- Source checked
- a2d149f · 2026-09-24
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
- 3GitHub stars
- Forks
- 0
- Open issues
- 0
- Last commit
- 2026-09-21
- Last release
- No release detected
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
Persistent semantic memory layer for DeepSeek Harness backed by Markdown and Milvus.
View plugin →LLM-supervised persistent memory and graph-based recall integration bundle for DeepSeek Harness.
View plugin →Lingshu long-term auditable memory system and cognitive graph for DeepSeek Harness.
View plugin →Composable, view-based layered memory bundle for DeepSeek Harness with pluggable providers and strategies.
View plugin →