Run DeepSeek-oriented terminal coding workflows with Reasonix
Use Reasonix when an operator wants a persistent terminal coding agent tuned for DeepSeek-style prefix caching, project memory, MCP-compatible plugins, and reviewable code-edit runs.
npx skills add agentskillexchange/skills --skill run-deepseek-oriented-terminal-coding-workflows-with-reasonix
npm i -g reasonix or brew install esengine/reasonix/reasonix, run reasonix setup, set the required provider API key environment variable, then use reasonix chat or reasonix run "<task>" inside the target project.Reasonix is a DeepSeek-native terminal coding agent distributed as a single Go binary through npm, Homebrew, and GitHub releases. It is useful when an operator wants an agent to stay in a project workspace, keep cache-stable context, call built-in or MCP-compatible tools, and run code tasks from the terminal rather than through a hosted chat surface.
What this skill actually does
Invoke this skill when the task is a repeatable engineering run: configure providers in `reasonix.toml`, keep secrets in environment variables, connect plugins or MCP-compatible subprocess tools, run `reasonix chat` for an interactive session, or use `reasonix run` for a bounded coding task. The scope boundary is the coding-agent operating loop: setup, provider/model configuration, tool/plugin wiring, sandbox and permission policy, task execution, and review of generated edits. It is not a generic DeepSeek listing or a broad CLI card.
Inputs and prerequisites: reasonix CLI, DeepSeek or OpenAI-compatible API key, optional MCP-compatible plugins.
Setup notes: Install with `npm i -g reasonix` or `brew install esengine/reasonix/reasonix`, run `reasonix setup`, set the required provider API key environment variable, then use `reasonix chat` or `reasonix run ““` inside the target project.
Source and verification boundary: use https://github.com/esengine/DeepSeek-Reasonix/blob/main/docs/GUIDE.md as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Custom Agents workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.