Adaptive Deep Study Coordinator Plugin
omdsh-dev/dsh-deep-research · skills
Adaptive Deep Study Coordinator plugin for DeepSeek Harness (official workflow engine, cybernetic/information theoretic design)
★ 21GitHub stars
0Forks
2026-08-15Last updated
TypeScriptLanguage
MITLicense
Key features
- TheoryImplementation in the plug-in
- The trigger relies on tool description (in-depth research/survey/comprehensive analysis of multi-source information, etc.).
- CyberneticsReference signal calibration (closed-loop control wrong target = in vain): The planning agent first defines the answer space (scope: what judgments/decisions the research supports) and the acceptance criteria (acceptance) of each sub-problem before starting research
- Ashby's law of necessary diversity (controller diversity system diversity ⇒ there must be blind spots): the planning agent enumerates the information dimensions of the topic, maps one dimension to each sub-problem, and outputs coverage self-check coverage_gaps
- Information TheoryInformation = Reduction of Uncertainty: Research sub-agents maintain three-state evidence confirmed / uncertain / gaps - an engineering expression of conditional entropy
Requirements
- This plug-in relies on DSH official workflow engine (ctx.workflows, peer: @deepseek-ai/dsh-workflow) when running.
- If the profile does not declare the provider (such as some Web Profile combinations), the Loader will remain pending - in this case, please register the workflows provider relationship in DSH Hub first or use a combination that provides this service instead.
Install command
dsh plugin --profile web add github:omdsh-dev/dsh-deep-research