AgentBible

Comprehensive development guidelines for AI coding agents. Works with Claude Code, GitHub Copilot, Windsurf, Cursor, and any tool supporting the AGENTS.md standard.

What's Inside

AgentBible/
├── AGENTS.md           # Universal agent instructions (always loaded)
├── rules/              # Always-on coding standards
├── skills/             # Language-specific knowledge (loaded on demand)
│   ├── python/
│   ├── csharp/
│   ├── cpp/
│   └── react/
├── workflows/          # Manual procedures (/command)
└── scripts/            # Deployment tools

Rules (Always On)

Cross-language principles loaded every session:

Rule	What it covers
clean-code.md	Naming, functions, comments, error handling, boundaries
patterns.md	Design patterns decision framework (when to use, when not to)
testing.md	Test pyramid, coverage, naming, mocking, anti-patterns
commit.md	Conventional commits, atomic changes, WHY not WHAT
architecture.md	ADR process, design principles, documentation standards
ipc.md	IPC mechanism selection (REST, gRPC, queues, shared memory)
mcp.md	Model Context Protocol integration, tool design, A2A communication
orchestration.md	Workflow patterns (Sequential, Concurrent, Handoff, Magentic)
security.md	Security guardrails, RBAC, sandboxing, audit trails, PII redaction
memory.md	Tiered memory architecture, decay, extraction, bi-temporal tracking
behaviour.md	Agent behavior, planning, Reflexion self-correction, circuit breakers

Skills (On Demand)

Language-specific knowledge that loads when you work with matching files:

Skill	Triggers on	Covers
python/	**/*.py	PEP 8, pytest, Ruff, async, patterns, IPC
csharp/	**/*.cs	Naming, xUnit, EF Core, ASP.NET, patterns
cpp/	**/*.cpp, **/*.hpp	Naming, Google Test, CMake, RAII, patterns
react/	**/*.jsx, **/*.tsx	Hooks, Jest, React Query, patterns

Workflows (Manual)

Multi-step procedures invoked with /command:

Command	What it does
/research	Structured research with plan confirmation and citations
/comment-guard	Verify comment truthfulness (WHY not WHAT)
/doc-drift-guard	Check docs match actual code and architecture
/adr-validator	Validate architecture decision records
/tech-debt-triage	Assess and prioritize technical debt
/static-analysis	Run language-appropriate static analysis tools
/evaluate	Analyze execution trajectories, detect failure patterns, self-correct

Quick Start

Option 1: Copy AGENTS.md

Copy AGENTS.md to your project root. Works with any agent that reads AGENTS.md files.

Option 2: Deploy to Specific Tools

# Deploy to GitHub Copilot
python scripts/deploy_to_copilot.py

# Deploy to Windsurf
python scripts/deploy_to_windsurf.py

# Deploy to Claude Code
python scripts/deploy_to_claude.py

Option 3: Use as Skills (Claude Code)

# Symlink into your project
ln -s /path/to/AgentBible/skills .claude/skills/agentbible

# Or copy specific skills
cp -r AgentBible/skills/python .claude/skills/python

How It Works

Category	When loaded	Purpose
AGENTS.md	Always	Universal instructions for any agent
rules/	Always (every session)	Cross-language coding standards
skills/	On demand (file match or manual)	Language-specific knowledge
workflows/	Manual only (/command)	Multi-step procedures with side effects

Supported Tools

Tool	How it uses AgentBible
Claude Code	AGENTS.md + rules/ as CLAUDE.md, skills/ as .claude/skills/, workflows/ as custom commands
GitHub Copilot	AGENTS.md as copilot-instructions.md, rules/ as .instructions.md
Windsurf	AGENTS.md + rules/ as .windsurf/rules/, skills/ as memories
Cursor	AGENTS.md as .cursor/rules, rules/ as .cursor/rules/*.md
Any agent	AGENTS.md in project root

Adding Content

New Language

1. Create skills/{language}/SKILL.md with frontmatter
2. Add topic files: clean-code.md, patterns.md, testing.md, static-analysis.md, parallel.md, ipc.md
3. Update deployment scripts if needed

New Rule

1. Create rules/{name}.md (under 200 lines)
2. Reference it in AGENTS.md

New Workflow

1. Create workflows/{name}.md with disable-model-invocation: true frontmatter
2. Reference it in AGENTS.md

See TEMPLATE.md for templates.

License

Internal use — Development guidelines for AI agents.