李永航
Community@bahayonghang
李永航 (bahayonghang) maintains 85+ skills spanning code auditing, academic writing, diagram generation, and knowledge-base governance for Claude Code and Codex environments.
Agent Skills by 李永航
Showing 84 vetted skills indexed across 5 GitHub repositories.
bahayonghangdrawio-academic-skills
Create publication-ready draw.io figures with academic policy gates and YAML-first workflows.
bahayonghanggh-pr-release
Inspect GitHub pull requests and publish releases with gh CLI.
bahayonghanghumanizer-paper
Remove AI-writing markers from academic manuscripts while preserving scholarly register norms.
bahayonghangdeep-research-pro
Perform multi-source web research to generate cited reports on complex topics.
bahayonghangroundtable
Simulate structured debates between real-world figures to uncover conceptual tensions.
bahayonghangliterature-mentor
Provide interactive, tutor-led analysis of academic papers using Zotero for retrieval.
bahayonghanggit-commit
Analyze staged changes and draft Conventional Commits with safety scans.
bahayonghangrenhua
Remove AI-flavored patterns from Chinese technical writing.
bahayonghangbidwriter
Analyze tender documents and generate compliant bid proposals.
bahayonghangtouying
Create animated Typst slide decks with themes and incremental reveals.
bahayonghangbeautiful-mermaid-editor
Modify TypeScript source and SVG post-processing in the Beautiful Mermaid live editor.
bahayonghangunknowns-first
Diagnose ambiguous tasks by identifying four unknown categories and success standards.
bahayonghangcode-auditor
Audit code across correctness, security, performance, readability, testing, and architecture.
bahayonghangcodex-bridge
Coordinate AI agent collaboration with Codex CLI for code implementation and review.
bahayonghangcodex-dynamic-workflows
Orchestrates complex multi-agent workflows by decomposing tasks into disjoint, verifiable packets with approval gates and integration verification.
bahayonghangrust-build-optimization
Diagnose slow Rust and Cargo builds using cargo --timings and linker analysis.
bahayonghangcode-refactor
Split modules, extract functions, and remove dead code while preserving behavior.
bahayonghangcode-quality-review
Review source code for maintainability, abstraction quality, and branching complexity.
bahayonghanghtml-artifact
Generate self-contained, accessible HTML artifacts for complex work outputs.
bahayonghangspark
Convert feature ideas into approved implementation plans with Markdown artifacts.
bahayonghangacademic-figure
Generate journal-compliant academic figures from raw data using Matplotlib, Seaborn, or Plotly.
bahayonghangast-grep
Search and validate code patterns using Tree-sitter structural queries.
bahayonghangclaude-context-improver
Audit Claude Code context files and propose optimization diffs.
bahayonghangcodex-workflow-recommender
Audit repository Codex configurations and recommend minimal workflow improvements.
Frequently Asked Questions About 李永航
FAQPage SchemaWhat tasks can I accomplish with bahayonghang's skill collection?▼
You can audit and refactor code, draft Conventional Commits, manage GitHub PRs and releases, generate draw.io/Mermaid/Excalidraw diagrams, edit PDF/DOCX/XLSX/PPTX files, polish LaTeX theses and IEEE papers, and run review-gated Obsidian knowledge bases.
Who is the target audience for these skills?▼
Software engineers using Claude Code or Codex CLI, academic researchers writing LaTeX/Typst papers, and technical writers maintaining documentation or Obsidian vaults. Rust, Vue, and Python developers get dedicated guidance skills for their stacks.
How do the drawio and drawio-academic-skills work together?▼
The drawio skill handles general diagrams via an offline YAML-first workflow requiring Node 20+. The academic overlay requires the sibling drawio folder and adds venue, color, caption, and readability gates for paper, thesis, and IEEE camera-ready figures.
Are these skills free and what license applies?▼
The new drawio skills are MIT licensed and hosted at github.com/bahayonghang/drawio-skills. They run on macOS, Linux, and Windows with Node 20+; draw.io Desktop is optional for 300dpi PNG export, otherwise falling back to standalone SVG.
What prerequisites do the knowledge-base skills require?▼
The kb-init, kb-ingest, kb-compile, kb-review, kb-query, and kb-render skills operate on an Obsidian vault with a raw/_manifest.yaml structure. The memory-system skill additionally builds a local SQLite vector and full-text index over Markdown notes.