README
dsh-terminal
An MIT-licensed persistent terminal mode for DeepSeek Harness. It keeps one real DSH Agent and Session alive across prompts, so customers can use Harness like a conversational coding CLI instead of only through the Web app or one-shot headless mode.
It reuses DSH's own model selection, tools, permissions, session log, commands, and shutdown lifecycle. There is no second agent implementation.
Install
dsh plugin --profile terminal add github:dongsheng123132/dsh-terminal
dsh --profile terminal
Plain JavaScript is committed as the package entry point, so Git installs require no build-script permission.
Commands
- ordinary text starts another turn in the same session
/newstarts a fresh session/helplists terminal commands and DSH's registered slash commands/exitor/quitcloses the terminal- other slash commands are dispatched directly to DSH (for example
/compact,/plan, or/cachewhen provided bydsh-cache-stabilizer)
To add cache-prefix stabilization and measured cache statistics:
dsh plugin --profile terminal add github:dongsheng123132/dsh-cache-stabilizer
Development
npm test
npm run check
更多「命令行与终端」插件
mnemon
作者 mnemon-dev
LLM-supervised persistent memory for AI agents — graph-based recall, cross-session knowledge, single binary. Works with DeepSeek Harness, Claude Code, OpenClaw, and any agent runtime.
sivtr
作者 ariestar
A unified agent memory workspace for human and agent
phi
作者 pulseaiclub
a coding Agent from pi. ∞ providers, sub-agents, hashline edits, and a permission gate
caliper
作者 edonadei
Know if your agent skill actually works. A lightweight evaluation harness that tracks a success rate across Claude Code, Codex, Pi, and Hermes.
