Brand assets · Favicon (PNG) · Favicon (ICO)

Security × Backend × Data • Python / Node.js / Dart (Flutter)

lit.link • Linktree • Portfolio (GitHub Pages)

Canonical source: profile.json This section is generated automatically. Edit the canonical JSON instead of this README block.

Bioinformatics → Security / Backend engineer.
I design and build practical CLI and browser tools: mostly single-file, standard-library Python, with Node.js for JavaScript libraries and Dart/Flutter for multi-platform apps. Each tool is verified in CI (cross-browser tests, JUnit evidence) and shared across repositories. My background spans genomic data pipelines, network security tooling, and developer productivity utilities.

“I build practical tools and templates you can run today.”

• 2020 — Started Bioinformatics: Began working with genomic data pipelines and sequence analysis tools.
• 2021 — Python GUI Tooling: Built beginner-friendly GUI wrappers for network security tools (Nmap, Scapy).
• 2022 — Security / Backend Focus: Shifted focus toward security tooling and backend service development.
• 2023 — Multi-language Engineering: Expanding into Go and TypeScript for backend APIs and developer tooling.
• 2026– — Single-file Tools & Verified Workflows: Building standard-library Python tools shared across repositories, cross-browser and Flutter Web verification in CI, LLM/MCP tool-call experiments, and question-driven baseball data analysis.

• Python (90%)
• Go (65%)
• TypeScript (60%)
• Bash (70%)

• Bioinformatics (85%)
• Network Security (75%)
• Backend / API (65%)
• GUI / Desktop (80%)

• Linux / CLI (85%)
• Docker (60%)
• Git / GitHub (80%)
• Wireshark / Scapy / Ghidra (70%)
• Flutter / Dart (60%)

• Single-file, standard-library Python tools shared across repositories (gh_identity, xprobe, yourself, cli_args, markdown)
• Question-driven baseball data analysis (Aoi and its subprojects)
• Cross-browser and Flutter Web verification with Playwright and CI evidence (browser-test-kit, flutter_navigation_basic)
• LLM / MCP tool-call experiments (mcp-toolcall-lab) and document conversion (Ironmate)
• DNA / protein sequence search and analysis (search_seq_including_spaces)

Portfolio • GitHub • Twitter • lit.link • Linktree

A diagram summarizing the flow of activity and the value it aims for. "This site" in the diagram refers to the portfolio. Click the image to enlarge it.

Selection source: projects.json · Repository facts: api/repos.json

Standard-library Python CLI for DNA and protein sequences: finds patterns whose front and back are fixed while the spacer in between varies, and writes CSV/JSONL. Also reads FASTA/FASTQ/GenBank, compares and translates sequences, gives RNA-fold and protein hints, and fetches from NCBI/PDB. CI-tested on Python 3.10–3.14.

bioinformatics dna protein cli

Baseball (NPB) research framework that looks for better questions rather than answers: starting from unexplained long-term outcomes, it observes, measures and tests the limits of models, metrics and evaluation. Parent of the six research subprojects below (git subtree).

baseball research statistics

Reproducible, polite fetching of NPB pages with conditional-GET caching and provenance, used to build observation datasets for Aoi.

baseball data-acquisition reproducibility

Observes and extracts the HTML/CSS/DOM structure of NPB pages before any interpretation (observe first, interpret later).

baseball html data-extraction

Team-season metrics, remaining-game and clinch magic-number simulation, and synthetic-league experiments from NPB game records.

baseball simulation statistics

Treats explanations as propositions and tests them against counterexamples, with rule search and sensitivity analysis for NPB outcomes.

baseball hypothesis-testing rule-search

Renders research records as Plotly.js HTML reports and proposition tables, with optional headless-Chrome PNG/SVG export and no Python plotting libraries.

baseball visualization plotly

Defines a research pipeline once in TOML and runs it with the same standard-library command locally and on GitHub Actions, with run receipts, Markdown link checks and publication gates.

pipeline github-actions reproducibility

Single-file, standard-library Python interface over the GitHub CLI for observing repositories, pull requests, checks and workflow runs, with explicit identity and bounded output.

github cli single-file

Single-file, standard-library toolkit for text search, boundary-case probing and failure-oriented test support, including convention-driven discovery of files and production symbols.

testing cli single-file

Single-file, standard-library tool that reports host, OS and runtime facts read-only, as concise copyable text for troubleshooting and AI context.

diagnostics cli single-file

Single-file, standard-library toolkit for reusable CLI argument definitions, validation, command construction, execution and JSON/text output.

cli validation single-file

Lightweight, function-focused Markdown utilities built with the Python standard library only.

markdown library single-file

Python utilities for ASCII art generation, conversion, layout and text-based diagrams. CI-tested on Python 3.10–3.14.

ascii-art diagram single-file

Error-triggered triage for Python, Node and CI failures: does nothing until something fails, then prints FOUND / NOT FOUND / HINT reports. Python standard library only.

debugging ci cli

Single-file, standard-library client for the NVD CVE API 2.0: fetches known vulnerabilities for CPE 2.3 names and emits deterministic JSON Lines. CI-tested on Python 3.9–3.14.

security cve cli

Cross-browser testing reference kit: Playwright Node and Python lanes on Chromium, Firefox and WebKit (desktop and mobile), screenshot evidence and OS/browser matrices on GitHub Actions, plus standard-library engineering utilities.

playwright cross-browser testing

Shared, build-free CSS themes, UI components and small JavaScript utilities for the static pages of these projects, including layout presets that switch without reloading.

css ui-components static-site

Multi-platform Flutter app (Web, iOS, Android, macOS) with 198 routed screens on a reusable foundation (themes, routes, shared widgets, URL parameters); the Flutter Web build is published on GitHub Pages and checked with Playwright screenshots.

flutter dart multi-platform

Reproducible experiments on how LLMs discover and call MCP tools: a FastMCP server with deterministic mock tools, Open WebUI and LibreChat as clients under test, protocol tests and an anti-pattern catalog.

llm mcp open-webui

Python toolkit for converting documents and images: a Markdown-to-PPTX pipeline, text and color editing of existing PPTX, browser-based Image Studio and Slide Studio, and pixel-grid art prototypes.

pptx markdown image

😏 詳細はこちら → myon-bioinformatics.github.io

gh_ops.py is the canonical stdlib-only GitHub operation CLI migrated from browser-test-kit, with its unit/parity tests owned here. See the contract. Consumers enroll the source/LICENSE via vendor_sync and keep only integration checks, rather than maintain generic operation implementations or duplicate tests.

For coordinated changes across repositories, fleet-apply --manifest fleet.json groups files per repository, reuses existing PRs and verifies explicitly listed source PRs before commenting/closing them. pr-edit --append-body-file updates metadata without discarding existing text; git-push sends an existing commit with destination readback. See the operation guide.

Immutable checkout verification reuses post-merge ancestry checks and ordered test commands across repositories, recording the tested commit/tree in a JSON receipt from a disposable detached worktree.

This repository also owns reusable CI workflows and stdlib-only tooling for sibling repositories. It is both the profile README and the parent tooling repo.

The Python runtime/setup policy separates stdlib runtime compatibility from development tooling and records the governed template boundary.

uv is the intended management path for these Python tools, including stdlib-only single-file tools. A tool is not kept off uv because someone also runs the file directly. When a standalone copy is what you want, copy the file from the GitHub repo (the repository and source path in vendor-catalog.json). Environments where uv does not apply, such as python -S or Python 3.9, are skipped with a reason. See the migration note.

The parent's Python tests workflow collects JUnit from four development/test Python versions (3.11–3.14) and three vendor-portability jobs (Linux 3.11, Windows 3.12, macOS 3.12). A separate stdlib-only runtime smoke job runs tests/stdlib_runtime_smoke.py with python -S on Python 3.9 and 3.10, without installing pytest. Raw XML and compact failure identity are separate artifacts: raw XML is retained for 14 days, while compact artifacts use repository-default retention. A complete collection does not turn a failed producer job green, and a green run may have an empty failure corpus.

Browser evidence currently provides a contract only. Shared Playwright and Stagehand execution adapters are not implemented here yet; consumer capture runners must remain until their replacement is verified. Repository metadata's contract/generator pair is still maintained in Ironmate.

The organization-wide target survey is in docs/vendor-target-inventory.json. Single-file candidates are catalogued even when they require an explicit skip: default_enrollment records a machine-readable default for every consumer; individual consumers entries override it. When absent, the original default_single_file_policy enrollment remains unchanged. Archived applications, external-package applications and parent-managed vendor tools are not implicitly placed into consumer trees. See catalog coverage.

Convention-driven file and symbol discovery gives verified one-liners for this repository. Supply filename/name rules to inspect production code, configuration and workflows with the shared stdlib-only xprobe tool.

Cross-repository use cases and connection candidates records the entry points and verification status of the 23 public, non-archived repositories.

Local prerequisite observations: python -S preflight.py --root . --require git --python-min 3.11. Optional --runtime-facts connects yourself's OS/runtime JSON without importing another checkout; see preflight usage and limits.