6 DeepSeek Harness Plugins to Cut Tokens & Streamline Context

Discover six DeepSeek Harness plugins that reduce token usage, manage context, and boost AI agent capabilities.

cpaua2 min16

Six plugins for DeepSeek Harness that help reduce token usage and tidy up the context.

Some of them handle data collection and storage, others extend the capabilities of AI agents, and some allow you to control their operation. Each solves a specific task, so it’s worth understanding its purpose before installing.

1. dsh-our-free-model — 49.7k stars

Allows the use of modern models without registration, authentication, or entering API keys. Just select the desired model. Local port forwarding is also supported.

The free pool has request‑rate limits, so transmitting confidential data through it is not recommended.

2. OpenViking — 41.1k stars

Organizes memory, knowledge, and Skills as a file system. Instead of loading the entire context, the agent receives short summaries and extracts information only from the necessary sections.

The claimed memory accuracy is 80–83 %, and token savings range from 34 % to 91 %.

The service requires prior configuration. Initial setup takes about 30 minutes.

3. Reasonix — 36k stars

An open AI agent for programming that connects to DeepSeek Harness via ACP as a sub‑agent.

It can analyze source code, edit files, run tests, and revert changes.

A two‑level configuration is provided. Note that when the plugin is disabled, the shell may simulate action execution.

4. routing-suite — 7k stars

Automatically selects a model based on the given task. It first classifies the request, then routes it to the appropriate model.

The claimed routing accuracy is 96 %, with a cache hit rate of 92–94 %.

Installation occurs in three stages and the plugin imposes strict version requirements on the core system.

5. dsh-ops — 4.723k stars

Reduces the size of returned results and hides some built‑in tools. PDF and Word documents are loaded on demand.

This prevents system instructions from being overloaded with unnecessary information, reducing token consumption.

However, the claimed savings have not yet been independently verified by third‑party developers.

6. dsh-context — 3.823k stars

Share

Author

cpaua

VibeCode blog admin. Writing about vibe coding, AI and open source.

Comments

To leave a comment, log in or sign up
Loading...

Weekly digest

The best of vibe coding, AI agents and open source — once a week, no spam.