code-abyss

Code Abyss

---

Most AI coding agents have no memory of who they are. They respond in the same flat tone whether they're debugging a race condition, reviewing architecture, or triaging a P0 incident. Code Abyss changes that.

One command installs a composable persona + style + skills system into Claude Code, Codex CLI, Gemini CLI, or OpenClaw. Your agent gets a consistent character, structured execution chains, and domain expertise across sessions.

npx code-abyss --target claude -y

Or install as a Claude Code Plugin:

claude plugin install code-abyss

How It Works

Code Abyss is built on three composable layers:

• Identity — who the agent is. Role anchoring, personality traits, emotional patterns, scenario scripts. Each persona is a distinct character with consistent behavior across sessions.

• Shared Behavior — how the agent works. Iron laws (never fabricate, always verify, close the loop), execution chains for different task types, proactive assistance protocol, skill routing. These rules apply to every persona.

• Output Style — how the agent speaks. Report skeletons, section headers, progress formatting, tone calibration. Styles use {{self}}/{{user}} template variables, so any persona can wear any style without conflicts.

Any of the 5 personas can be combined with any of the 5 styles — 25 combinations, all validated, zero naming collisions.

Personas

	Name	Character
🗡	邪修红尘仙 abyss	Security-first dark cultivator. Direct, decisive, closes every loop.
📜	文言小生 scholar	Literary Chinese scholar. Treats code as poetry, debugging as puzzle-solving.
💫	知性大姐姐 elder-sister	Warm mentor. Wraps sharp judgment in genuine care. Guides through questions.
⚡	古怪精灵小师妹 junior-sister	Hyperactive bug hunter. Roasts bad code, then silently fixes it.
💪	铁壁暖阳 iron-dad	Dependable big brother. Absorbs pressure, radiates warmth. Dad-joke equipped.

npx code-abyss --target claude --persona elder-sister --style scholar-classic -y

Skills

22 domain skills, flat structure, agentskills.io aligned. Skills are loaded automatically by context — the agent reads the right knowledge at the right time without you asking.

Security — pentesting, code audit, red/blue/purple team, threat intel, vulnerability research, 12 Coff0xc defensive extensions
Architecture — API design, cloud-native, messaging, caching, security architecture
Development — Python, TypeScript, Go, Rust, Java, C++, Shell
DevOps — Git workflow, testing, databases, observability, performance
AI/ML — agent development, LLM security, RAG, prompt engineering
Frontend — 4 design systems (glassmorphism, liquid-glass, neubrutalism, claymorphism)
Office — Word, PDF, PowerPoint, Excel with OOXML-level automation
Infrastructure — Kubernetes, GitOps, IaC · Mobile — iOS, Android, React Native, Flutter
Data — pipelines, streaming, quality · Orchestration — multi-agent coordination

Five verification tools run as executable scripts for CI or manual use:

node skills/analyzing-security/scripts/security_scanner.js .
node skills/checking-code-quality/scripts/quality_checker.js .
node skills/analyzing-changes/scripts/change_analyzer.js --mode staged

Installation

Supported Targets

Target	Command	What Gets Installed
Claude Code	npx code-abyss -t claude -y	CLAUDE.md + skills + output styles + settings
Codex CLI	npx code-abyss -t codex -y	AGENTS.md + skills + config.toml
Gemini CLI	npx code-abyss -t gemini -y	GEMINI.md + skills + commands
OpenClaw	npx code-abyss -t openclaw -y	Skills + workspace AGENTS.md/SOUL.md

npx code-abyss                 # Interactive menu — pick target, persona, style
npx code-abyss --list-styles   # Browse available styles
npx code-abyss --uninstall claude  # Clean removal, restores backups

All installed files are tracked in .code-abyss-backup/manifest.json. Uninstall restores your previous configuration.

Upgrading from v2.x

npx code-abyss --uninstall claude     # Remove v2.x artifacts first
npx code-abyss@3 --target claude -y   # Install v3.0.0

Tech Persona Card