mindspace-dsh-local-rag
A local-only hybrid RAG plugin for DeepSeek Harness that gives models an explicit tool for searching uploaded documents and session compaction summaries.
Install
$ corepack pnpm dsh plugin --profile web add $ragTgzPlugin 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
- Registers an explicit `search_local_memory` tool for model-directed retrieval.
- Combines vector similarity and BM25+ lexical search with reciprocal-rank fusion.
- Indexes uploaded PDF, DOCX, TSV, CSV, TXT, Markdown, JSON, and HTML content with source provenance.
- Indexes successfully committed native DSH `compaction/end` summaries in a session-isolated corpus.
- Supports document and summary viewing, editing, logical deletion, restoration, and immutable revisions.
- Provides source-specific follow-up through `local-rag://source/...` addresses without exposing filesystem paths.
- Offers model download, integrity verification, explicit start/stop, auto-start, index health, and rebuild controls.
- Keeps lexical retrieval available when the vector runtime is unavailable or stopped.
Useful For
- Search locally uploaded reference documents during a DeepSeek Harness conversation.
- Retrieve facts from older, compressed summaries within the current session.
- Import mixed office, PDF, tabular, and text-based knowledge sources for local lookup.
- Use lexical search before an embedding model has been started.
- Review retrieval provenance and perform follow-up searches limited to a document or source.
Who It Fits
- DeepSeek Harness users who need local document retrieval.
- Users working with private or offline knowledge collections.
- Developers integrating tool-calling models in a local DSH profile.
- Teams that need session-isolated retrieval over DSH compaction summaries.
Documented Limitations
- Requires a local DeepSeek Harness checkout, Node.js 22.19+ or 24+, and pnpm for the documented installation flow.
- The embedding model runtime must be downloaded and explicitly started for vector retrieval; lexical retrieval can operate without it.
- Network access is required to download pinned model artifacts.
- Retrieval uses fixed candidate and output limits; the model cannot alter ranking limits.
- The plugin intentionally does not include reranking or user-configurable Top-K.
- A DSH profile must enable exactly one structured-memory implementation providing the `sessionMemory` service.
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 6 runtime, 9 development, 6 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
- Spirtxiaoqi7/mindspace-dsh-local-rag
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin
- Source checked
- 1da456b · 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
- 2GitHub stars
- Forks
- 0
- Open issues
- 0
- Last commit
- 2026-08-21
- Last release
- 2026-08-14
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
Persistent multi-model Agent teams and bounded DAG workflows with GUI controls for DeepSeek Harness.
View plugin →Second-model AI auto-review for DeepSeek Harness approval requests with fail-closed safety and session audit.
View plugin →Native SQLite-backed local project taskboard plugin and UI for DeepSeek Harness with Agent claim and review flows.
View plugin →Preview, create, and edit spreadsheets, docs, slides, and canvases in DeepSeek Harness powered by Univer.
View plugin →