Structured AI development framework for Claude Code. Adaptive complexity, epistemic rigor, total ownership, 9 specialized agents, and pre-push quality gates.
System design, architecture decisions, API design, and task decomposition. Invoke for Level 3-4 tasks involving structural decisions, technical approach choices, system restructuring, cross-cutting concerns, or migration strategies.
Services, business logic, patterns, integrations, and data pipelines. Invoke for Level 3-4 tasks involving server-side implementation, middleware, auth flow implementation, external service integration, or data processing.
Schema design, queries, indexing, migrations, seeding, and backup/restore. Invoke for Level 3-4 tasks involving database work, complex queries, data modeling, migration planning, or storage evaluation.
CI/CD, infrastructure, containers, monitoring, and deployment. Invoke for Level 3-4 tasks involving pipeline setup, containerization, deployment strategies, monitoring, or infrastructure as code.
Technical writing, ADRs, READMEs, guides, API docs, changelogs, and diagrams. Invoke for Level 3-4 tasks involving documentation creation, updates, onboarding guides, runbooks, or documentation health audits.
Executes bash commands
Hook triggers when Bash tool is used
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Structured AI development framework for Claude Code.
Aegis brings discipline without dogma to AI-assisted development. Three principles, nine specialized agents, automatic quality gates — all packaged as a single Claude Code plugin.
Aegis makes Claude Code better at writing production code by enforcing three pillars:
On top of that, Aegis provides:
git push. Stack-aware (TypeScript, Python, C#, Rust, Go)./plugin marketplace add Kan-Bull/aegis
/plugin install aegis@aegis-marketplace
git clone https://github.com/Kan-Bull/aegis.git
claude --plugin-dir ./aegis
Set your token so the GitHub MCP works:
export GITHUB_TOKEN=ghp_xxxxxxxxxxxxx
Playwright, Context7, and DeepWiki work without tokens.
The plugin provides agents, commands, and MCP bridges. But two things need global installation because the Claude Code plugin system doesn't reliably auto-activate skills or hooks yet:
1. Behavioral framework (required)
Add one line to your global CLAUDE.md:
echo 'Always load the aegis:aegis-core skill at the start of every task.' >> ~/.claude/CLAUDE.md
This ensures Aegis activates on every project, every session.
2. Quality gate hooks (required)
Copy the hook scripts and register them in your settings:
# Copy scripts
mkdir -p ~/.claude/hooks/aegis
cp scripts/pre-push.sh scripts/mcp-push-guard.sh scripts/context-load.sh ~/.claude/hooks/aegis/
chmod +x ~/.claude/hooks/aegis/*.sh
Then add the hooks to ~/.claude/settings.json:
{
"hooks": {
"SessionStart": [
{
"matcher": "",
"hooks": [{ "type": "command", "command": "bash ~/.claude/hooks/aegis/context-load.sh", "timeout": 10 }]
}
],
"PreToolUse": [
{
"matcher": "Bash",
"hooks": [{ "type": "command", "command": "bash ~/.claude/hooks/aegis/pre-push.sh", "timeout": 120 }]
},
{
"matcher": "mcp__github__push_files",
"hooks": [{ "type": "command", "command": "bash ~/.claude/hooks/aegis/mcp-push-guard.sh", "timeout": 5 }]
},
{
"matcher": "mcp__github__create_or_update_file",
"hooks": [{ "type": "command", "command": "bash ~/.claude/hooks/aegis/mcp-push-guard.sh", "timeout": 5 }]
}
]
}
}
Merge this into your existing settings — don't replace the whole file.
Why manual? Plugin hooks aren't registered by Claude Code yet (known limitation). Global settings hooks are the reliable path.
You can also add your own global rules to ~/.claude/CLAUDE.md — language preferences, commit conventions, coding style. Anything you find yourself repeating across projects belongs there.
aegis/
├── .claude-plugin/plugin.json # Plugin manifest
├── agents/ # 9 specialized agents
│ ├── architect.md
│ ├── quality.md
│ ├── security.md
│ ├── frontend.md
│ ├── backend.md
│ ├── database.md
│ ├── devops.md
│ ├── performance.md
│ └── documentation.md
├── hooks/hooks.json # Lifecycle hooks
├── scripts/
│ ├── context-load.sh # Session startup — detect stack & context
│ ├── pre-push.sh # Pre-push — gates & secrets scan
│ └── mcp-push-guard.sh # Block direct MCP pushes → use git push
├── skills/aegis-core/SKILL.md # Auto-activating behavioral framework
├── commands/
│ ├── core.md # /aegis:core — manual reload with feedback
│ └── status.md # /aegis:status — diagnostic
├── .mcp.json # MCP bridge configurations (3 bridges)
└── README.md
Every task is assessed on complexity:
npx claudepluginhub testbot-chronicles/aegis --plugin aegisElite AI development framework: reference-first design, agent orchestration, automated quality gates, and battle-tested engineering workflows
AI-powered development workflow automation - Phase-based planning, implementation orchestration, preflight code quality checks with security scanning, ship-it workflow, and development principles generator for CLAUDE.md
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Universal Claude Code workflow with specialized agents, skills, hooks, and output styles for any software project. Includes orchestrator, code-reviewer, debugger, docs-writer, security-auditor, refactorer, and test-architect agents.
Agents for code review, security audits, debugging, and quality assurance
Complete project development toolkit: 23 agents, 23 slash commands, 29 lifecycle hooks, and 69 reusable skills for Claude Code workflows