Qiongli

English

简体中文

English

简体中文

Appearance

QiongliUse AI agents for academic research without losing the evidence trail.

Turn a topic into a paper route, task IDs, quality gates, literature and citation evidence, writing and code outputs, and review handoffs.

Quickstart
Install
Choose A Workflow

Start Small

Use a native plugin, Desktop ZIP, bootstrap, npm, or pipx path based on the task instead of installing every runtime up front.

Route The Work

Map research goals to paper types, stages, Task IDs, expected outputs, and quality gates.

Keep Evidence Visible

Track claims, citations, search logs, diagnostics, methods, code, review status, and handoffs under predictable paths.

Separate Update From Refresh

`qiongli update` upgrades the package; `qiongli upgrade` refreshes local assets. The boundary is explicit.

Choose Your Entry Point ​

You want to...Start here
Try Qiongli in one clientInstall
Get from no setup to a first workspaceQuickstart
Know what to type after installUsing Agent Skills
Choose a paper workflowTask Recipes
Run validators, doctor, or orchestrated tasksMulti-Agent Runtime
Automate installs, checks, updates, or release workCLI Reference

Latest Stable Downloads ​

Current stable release: v1.17.0. These direct links cover the common install paths; use the download guide for subject-specific Desktop ZIPs and maintainer artifacts.

NeedLink or command
npm CLIqiongli@1.17.0: npm install -g qiongli@latest
PyPI CLIqiongli 1.17.0: pipx install qiongli
Claude Desktop recommended pluginqiongli-claude-desktop-plugin-v1.17.0.zip
Claude Desktop/Web fallback skill ZIPqiongli-claude-desktop-skill-core-v1.17.0.zip
Claude Desktop literature MCPBqiongli-literature-provider-0.1.5.mcpb
Zotero Desktop companionqiongli-zotero-companion-0.2.2.xpi
All release assetsDownload guide and GitHub Release

What The System Covers ​

Qiongli ships the portable qiongli-workflow package plus optional local runtimes for literature search and orchestration.

  • Framing: questions, gaps, contribution claims, venues, and boundaries.
  • Literature: provider-aware searches, diagnostics, bundles, screening, extraction, and snowballing.
  • Design: variables, datasets, robustness, preregistration, ethics, and data management.
  • Writing: claim-evidence mapping, tables, figures, limitations, proofreading, submission, and rebuttal.
  • Code: Stage-I specification, planning, execution, and review for methods-heavy work.
  • Coordination: solo, duo, or triad roles across local agent tools with recorded handoffs.

Runtime Boundary ​

Installing workflow assets does not imply local agent execution. You can use Qiongli as a skill/plugin without Python. Python 3.12+, model CLIs, and matching authentication are only required for doctor, validators, MCP orchestration, or actual task execution.

Documentation Map ​

  • Guide: install, usage, upgrades, troubleshooting, and runtime choices.
  • Examples: paper-type playbooks.
  • Reference: CLI behavior and skill catalog.
  • Architecture: package surfaces, contracts, roles, and bridges.
  • Advanced: MCP providers, Zotero, subject packaging, and plugin-first distribution.
  • Maintainer: release policy, naming policy, and contributor guidance.

Qiongli documentation

MIT License