dsh-tokensave
Run the following command in DeepSeek Harness:
dsh plugin install Miku196/dsh-tokensave
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install Miku196/dsh-tokensave in your DeepSeek Harness terminal to install; source code is available at https://github.com/Miku196/dsh-tokensave .
About this plugin
When AI coding agents explore a codebase, the most token-hungry move is repeated grep and whole-file reads, where the useful signal is only a fraction of what gets pulled in and the rest is pure waste. dsh-tokensave plugs the tokensave Rust semantic-graph engine directly into DeepSeek Harness, letting agents issue structured semantic queries instead of blind text scans and saving 900-2000 tokens per lookup.
Once installed, the agent gains 80+ graph tools (search, callers, callees, body, entities, dead_code, circular, and more) with system-prompt guidance to prefer them for code exploration. Indexing is fully zero-config: every call follows the current session working directory, walks up to the nearest existing index, auto-inits in a brand-new folder, and never auto-indexes the home directory or its direct children to avoid accidental whole-home scans. A missing binary degrades gracefully without blocking profile startup.
Built for DSH users working on large codebases who want to cut code-exploration token costs significantly. Install and go, no config file edits required.
Use Cases
- Replace grep and whole-file reads with semantic queries when agents explore large codebases
- Trace caller/callee relationships without reading through source files
- Auto-build and reuse a semantic index across subdirectories with zero configuration
Best For
- Developers using DSH for daily coding tasks
- Teams focused on reducing agent token consumption and cost
- Engineers working in large monorepos or polyglot projects
Related Plugins
Traceable, searchable cross-session memory for AI agents that turns conversation knowledge into a typed knowledge graph and recalls relevant subgraphs instead of replaying full history, natively integrated with DeepSeek Harness.
Gives DSH AI cross-session long-term memory, to-do and skill management, plus multi-session orchestration, external AI delegation, and an infinite canvas that grows with you.
dsh-mnemon is a three-tier, pluggable, Agent-driven memory system for DeepSeek Harness, combining Runtime memory, Project Documents, and replaceable Memory Spaces with nine long-term providers.
An opinionated, zero-infrastructure file-based memory protocol: human-readable markdown files plus a bounded always-loaded index, curation discipline, and no database or embeddings.