Skip to content
SSCodeAI
日本語English

Resume

SSCodeAI's public project resume — open source projects, technical focus, and public artifacts.

Updated View as Markdown

Resume

SSCodeAI’s public resume is organized around public projects and artifacts rather than personal information.

Summary

SSCodeAI maintains AI coding workflows and agent libraries for Japanese IT delivery. Building on upstream open-source projects, it performs local adaptation with consideration for Japanese teams, acceptance, audit trails, incident reporting, handover, and compliance.

Public Projects

Project Role Public output
superpowers-ja Japanese workflow adaptation and maintenance skill catalog, tool install docs, audit scripts, eval transcripts, release notes
agency-agents-ja Japanese agent library adaptation and expansion 411 agents, 27 workflows, GIS / healthcare / gov-tech / platform engineering / research / document generation / China networking coverage, timely-disclosure / TDnet compliance reviewer, translation progress, upstream coverage
sscodeai-site Documentation and publication site docs, tutorials, blog, generated project catalogs, Cloudflare deploy docs
takumi Verifiable agentic software delivery platform workflow engine, runtime adapters, traceability, evidence-native testing, Agent Eval, MEA loop, Japanese SI / V-model workflow
aiitg Input-side zero-trust security layer 7 detectors × 6 formats, sanitization, trust labeling, policy enforcement, MCP server, Claude Code hook, 191 tests
ai-skill-maker Reusable AI skill generation meta-skill renderer / validator scripts, Trust Gate, release gates, 9,000-token file budget, fingerprint lock on protected core principles
ai-project-skill-maker Project maintainer skill generation meta-skill genesis / repo modes, evidence model, adapters for AGENTS.md through Copilot, forward-test prompts
skills-ja-JP Japanese localization and maintenance of Agent Skills 20 skills (15 engineering + 5 productivity), translation policy, translation / audit scripts, release record
decision-workbench Decision-space expansion tool six-stage pipeline, FastAPI + React, hidden-gold eval (26% → 78% → 100%), 25 tests
decision-space-expansion Decision methodology as a skill Agent Skills–format SKILL.md, three-stage loop, per-tool install locations, pitfalls
eli5 Picture-first explainer generator SKILL.md spec, HTML template, PDF / PNG export script, en / ja / zh examples
fusion-cache LLM API caching layer L1 / L2 / L3 layers, benchmark (97% hit rate, P95 0.7 ms), gateway, MCP server, Redis store

Technical Focus

  • AI coding agent workflow design
  • Prompt / skill localization for Japanese development teams
  • Agent role design for SIer, custom development, in-house services, SaaS, EC, manufacturing DX, and public sector
  • Documentation automation with Astro / Nimbus / MDX
  • Release hygiene: validation scripts, generated docs sync, version notes, quality reports

Representative Work

superpowers-ja

  • Install docs for multiple tools (Claude Code, Cursor, Codex CLI, Gemini CLI, OpenCode)
  • Japanese workflows for code review, incident reporting, acceptance-test specs, Git operations
  • Upstream / Chinese-edition diffs documented in version differences
  • Release quality verified with audit scripts and eval transcripts

agency-agents-ja

  • Agent roles adapted for the Japanese market, based on upstream agency-agents
  • Localized ideas from the Chinese edition (agency-agents-zh) for IoT / financial anti-fraud as Japan-market agents
  • source, translation_status, upstream_path in frontmatter, synced to generated docs
  • Upstream / zh attribution and Japan-market diffs documented in README and coverage reports

sscodeai-site

  • Auto-generates docs from superpowers-ja and agency-agents-ja
  • Publishes agent catalog, workflow catalog, translation progress, and upstream coverage
  • Static build and sitemap for Cloudflare Pages

takumi

  • Swaps fake / Pi / DeepSeek / CLI bridge runtimes through a model-agnostic adapter API
  • Connects requirements, design, tests, evidence, and review through a traceability matrix
  • Strengthens verify-before-done delivery with Agent Eval, held-out verifier tests, and a Manage-Execute-Audit loop

aiitg

  • A 7-detector × 6-format chain with coordinate-level JSON evidence makes hidden prompt injection auditable
  • Sanitize / trust / policy stages apply strip or redact, safe-to-dangerous labels, and allow-to-block decisions
  • An MCP server and a Claude Code PreToolUse hook (fail-closed, cached, audited) give agents zero-touch adoption

ai-skill-maker

  • Renders functional / document / workflow / project maintainer skills and adapter bundles from a config
  • A Trust Gate plus evidence labels separate facts, intent, standards, and assumptions
  • A SHA-256 fingerprint lock on protected core principles and a 9,000-token file budget prevent drift on refresh

ai-project-skill-maker

  • Genesis and repo modes generate maintainer skills while keeping declared intent separate from observed repository facts
  • The evidence model (observed_fact / declared_intent / recommended_standard / inferred_assumption) applies across every generated reference
  • Adapters for AGENTS.md, CLAUDE.md, Cursor, and Copilot render from the same config

skills-ja-JP

  • Initialized the Japanese localization of 20 skills against upstream sscodeai/skills@272f99b (local commit 247525d)
  • Documented a translation policy that preserves directory names, skill names, commands, paths, URLs, and identifiers
  • check-translation.mjs and audit-english.mjs verify translation consistency and surface untranslated prose

decision-workbench

  • Exposes the six stages — bias flag, independent divergence, cross-exam, option map, constraints, human decision — as an API
  • Measures the improvement from single model 26% to 100% against a hidden gold option set
  • Built as a diverger with a mandatory human gate rather than an llm-council-style converger

decision-space-expansion

  • Distills the decision-workbench method into a single SKILL.md with no code and no dependencies
  • Structures divergence into consensus / true divergence / pseudo-divergence zones
  • Documents install locations for Claude Code, Cursor / Windsurf, Codex, and Gemini CLI

eli5

  • Defines a picture-first explainer format where each step is one large inline SVG plus a one-line caption
  • Achieves zero-external-request output through a single HTML template with a quiz and interactive glossary
  • Ships examples/what-is-api-{en,ja,zh}.html as the quality bar

fusion-cache

  • Fuses exact / semantic / prefix accounting into a single middleware
  • Recorded a 97% hit rate and P95 0.7 ms over 20 tasks × 3 reps (60 requests)
  • An OpenAI-compatible gateway and an MCP server make it usable from agents with zero code changes

Public Signals

Privacy Notes

This resume intentionally avoids private phone numbers, home address, customer names, contract details, and internal project information.

Navigation

Type to search…

↑↓ navigate↵ selectEsc close