DSH-EvoResearch
EvoResearch is a DeepSeek Harness plugin by Karbo123, described in its README as a self-evolving research workflow and an agent workbench spanning the research lifecycle: topic investigation, literature reading, experiment design and paper writing. The README describes research memory, Chat Graph branching, a project workspace with '@' references, an experiment console, multi-agent collaboration, scheduled research tasks and WYSIWYG writing, with data stored locally on the user's own computer. T
Installation
No reliable install command was detected. Check the project README for installation instructions.
Overview
EvoResearch is a DeepSeek Harness plugin by Karbo123, described in its README as a self-evolving research workflow and an agent workbench spanning the research lifecycle: topic investigation, literature reading, experiment design and paper writing. The README describes research memory, Chat Graph branching, a project workspace with '@' references, an experiment console, multi-agent collaboration, scheduled research tasks and WYSIWYG writing, with data stored locally on the user's own computer. T
- Research memory: important conclusions and research notes are described as retained for reuse in later conversations
- Chat Graph branching: fork different hypothesis routes from any message, advance them in parallel and trace back to key nodes
- Project workspace where '@' can reference workspace files or past sessions, read through the DSH official reference chain
- Experiment console combining ledger, rounds and daily reports, recording what each experiment round did and why
- Multi-agent collaboration with research roles such as planning, investigation, coding, analysis and writing
- Scheduled research tasks that can run literature tracking and progress summaries automatically
- WYSIWYG writing with real-time rendering of headings, tables, formulas and flowcharts
- README states data is local-first: sessions, memory, project files and experiment records stay on your own computer