dsh-long-term-memory
Deterministic, layered long-term memory for DeepSeek Harness with BM25 recall and JSONL storage.
Install
$ dsh plugin --profile web add .Plugin 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
- Three-tier storage hierarchy spanning user profiles, global facts, and project workspace memory
- CJK-aware BM25 keyword recall with deterministic and reproducible search scores
- Synchronous prompt context injection during request assembly (recent, full, or off modes)
- Built-in threat pattern scanning for prompt injections and malicious memory writes
- Model tools including memory_write, memory_recall, memory_list, memory_forget, memory_export, and memory_batch
- User slash commands (/memory) and web management UI in the DSH client details panel
- Atomic JSONL persistence with cross-process file locking and external drift detection
- Optional post-turn LLM auto-summarization and capacity-aware memory compression
Useful For
- Persisting user preferences, coding conventions, and persona details across distinct sessions
- Retaining architectural decisions and constraints scoped directly to a Git repository workspace
- Searching past conversation notes and facts deterministically without external vector embeddings
- Reviewing and editing accumulated agent memories through the DSH web UI or slash commands
Who It Fits
- DeepSeek Harness developers needing cross-session contextual continuity
- Users who want lightweight, reproducible local memory without vector DB infrastructure
- Teams tracking project-level decisions within workspace repository files
Documented Limitations
- Uses BM25 lexical matching rather than semantic vector embeddings
- Enforces configured character limits per memory store (defaults to 20,000 characters per scope)
- Requires manual restart of dsh web after installation
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 4 runtime, 0 development, 4 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.
GitHub reports this repository as archived.
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
- windrover/dsh-long-term-memory
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin
- Source checked
- 3740ba8 · 2026-08-30
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
- 2026-08-27
- Last release
- No release detected
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
Composable three-tier memory control plane for DeepSeek Harness with runtime context, searchable documents, and pluggable long-term memory.
View plugin →Long-term spatiotemporal memory, zero-LLM whitebox reasoning, and knowledge flywheel integration for DeepSeek Harness via AEIS.
View plugin →Traceable, searchable, cross-session knowledge graph memory engine for DeepSeek Harness.
View plugin →Import and resume chat histories from 18+ AI coding agents into DeepSeek Harness with reverse export and sync.
View plugin →