dsh-local-ai
Local-model (Ollama and OpenAI-compatible) integration and task-based routing with cloud fallback for DeepSeek Harness.
Install
$ dsh plugin --profile web add github:PerryLink/dsh-local-aiPlugin 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 Ollama and OpenAI-compatible LLM adapters via ctx.llm.registerAdapter
- Inspects, pulls, lists, and removes local models using dedicated tools (ollama_list, ollama_show, ollama_pull, ollama_remove)
- Provides a /ollama slash command and health check tool reporting both process and API responsiveness
- Rule-based routing to local models by task purpose, keywords, or default preference with automatic cloud fallback before first token output
- Optional vision capability negotiation when models report multimodal support
Useful For
- Routing sensitive or confidential prompts to local offline models while keeping standard tasks on cloud providers
- Managing and monitoring local Ollama model disk usage, running state, and health directly from DeepSeek Harness
- Delegating routine background tasks like session compaction and title generation to fast local models
Who It Fits
- Developers running local LLMs through Ollama, LM Studio, vLLM, or llama.cpp
- Privacy-conscious users wanting automatic offline routing for confidential tasks in DeepSeek Harness
- Power users managing local model inventories and fallbacks
Documented Limitations
- Only pre-token failures trigger automatic cloud fallback; failures that occur mid-stream after generation starts are forwarded to the caller
- Image input is supported only for models explicitly reporting vision capabilities via /api/show; text-only models reject multimodal payloads
- Requires an accessible Ollama or OpenAI-compatible backend service running locally or over the network
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 2 runtime, 22 development, 8 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.
package.json declares a prepare lifecycle script that may run during relevant package installation workflows.
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
- PerryLink/dsh-local-ai
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin
- Source checked
- db47771 · 2026-10-16
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
- 20GitHub stars
- Forks
- 1
- Open issues
- 1
- Last commit
- 2026-10-12
- Last release
- 2026-10-10
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
Persistent multi-model agent teams with dynamic DAG orchestration, run monitoring, and token insights for DeepSeek Harness.
View plugin →Compatibility bridge running unmodified Pi ecosystem extensions natively on DeepSeek Harness.
View plugin →Full office suite integration for DeepSeek Harness supporting spreadsheets, docs, slides, bases, and canvases.
View plugin →Theme, wallpaper, and styling plugin for DeepSeek Harness with 8 preset themes, glassmorphism, accent colors, and custom theme packs.
View plugin →