DSH OPS is an open-source toolkit that extends DeepSeek Harness with efficient file, document, and shell operations for AI agents. It is designed to reduce unnecessary token consumption and improve task execution workflows. The software uses a Rust-based worker to provide bounded file access, compact responses, and structured operation results. It supports reading and discovering files while preserving line numbers and limiting excessive output. Additional tools handle CSV, TSV, DOCX, and supported PDF text extraction operations. Developers can access functionality through native agent tools or a programmatic TypeScript interface. Its modular architecture supports independent component configuration, permission checks, and integration with desktop, web, and terminal-based DSH environments.
Features
- Token-Efficient AI Agent Operations
- Bounded File Reading and Discovery
- Rust-Based Execution Worker
- Structured Document and Data Extraction
- Cross-Platform Bash Command Integration
- Modular DSH Plugin Configuration