openJiuwen
Official@openjiuwen-ai · China
openJiuwen is an open-source community dedicated to building a production-ready AI Agent platform.
Agent Skills by openJiuwen
Showing 63 vetted skills indexed across 2 GitHub repositories.
gitcode
Operate GitCode issues, pull requests, wikis, and org templates via the gitcode CLI.
ci
Diagnose and modify CodeArts and GitHub Actions CI pipelines for the ScienceDiscovery repository.
deploy
Deploy and manage the ScienceDiscovery stack via host processes or Docker Compose.
e2e-testing
Design and run user-perspective E2E journeys for ScienceDiscovery via Playwright, HTTP API, or CLI.
create-pr
Open GitCode merge requests with local CI validation and result interpretation.
idea-tree-team
Prepares literature-grounded inputs and hands off autonomous Idea Tree research to the Python engine.
structure-pocket-inspection
Inspect PDB structures and identify protein residues near a ligand or pocket center.
evidence-extractor
Extracts structured, source-grounded evidence items from academic source packages for downstream synthesis.
creative-material-design
Generates water treatment material designs with structural descriptions, property predictions, and feasibility analysis.
code-engineer
Execute Python and R code to inspect, analyze, and export structured data results.
insight-aggregator
Synthesize multiple independent expert assessments into a concise cross-validated insight for tree propagation.
antibody-protenix-pipeline
Submit and monitor RFdiffusion-ProteinMPNN-Protenix antibody design jobs through the ScienceDiscovery NPU Broker.
report-writer
Synthesizes domain summaries into formatted research reports with cross-domain connections and verbatim contradictions.
science-research-team
Orchestrates multi-domain research teams for literature review, data analysis, and report synthesis.
literature-searcher
Retrieves and deduplicates academic papers from arXiv, PubMed, and CrossRef into structured JSON.
assessment-screening
Scores material design candidates against configurable multi-dimensional expert rubrics.
evolve-design
Designs and launches checkpointed evolutionary search runs that iteratively optimize text-shaped candidates against measurable scores.
result-evaluator
Evaluate analysis results for accuracy, reliability, and statistical rigor with structured verdicts.
life-science-evidence-brief
Generate cited evidence briefs from UniProt, PubMed, Europe PMC, and arXiv records.
citation-reviewer
Verify cited paper identity and check whether Artifact claims are supported by linked Evidence.
computation-reviewer
Verify numeric claims against linked Evidence entries in locked Artifacts.
skill-name-1
Analyzes delegated expert subtasks and returns evidence-backed conclusions with risks and next steps.
content-enhancement
Enhance existing drafts with citations, opening rewrites, title optimization, and text polishing.
content-drafting
Draft structured documents and chapters from scattered notes, outlines, and interview records.
Frequently Asked Questions About openJiuwen
FAQPage SchemaWhat specific tasks can I perform using these capabilities?▼
You can perform kernel profiling, convert SOP documentation into executable skill structures, extract line items from financial documents, and analyze MoE phase timing using Perfetto trace data. These functions enable systematic documentation auditing and hardware-specific performance optimization for complex computational models.
Which technical personas benefit most from this platform?▼
This platform is designed for machine learning engineers, kernel developers, and technical documentation specialists. It specifically supports those working on Ascend hardware optimization, MoE architecture tuning, and enterprise-grade knowledge base management through structured document ingestion and performance benchmarking.
What are the prerequisites for running these optimization tasks?▼
Execution requires a local environment configured for the JiuwenSwarm runtime, including access to the skvm environment for profiling and JIT optimization. Users must also ensure their source data, such as SOP documents or trace files, are formatted correctly for the respective ingestion and analysis modules.