AI Context Docs Lifecycle

Knowledge-first methodology kit for AI-assisted development

---

What Is This?

AI Context Docs Lifecycle is a knowledge-first methodology kit for AI-assisted development. It gives AI coding assistants structured context and workflows for understanding your project and building features consistently.

Pick the modules you need:

#	Module	Purpose
1	Foundation	AGENTS.md bootstrap + reference doc templates
2	Dev Workflow	Feature workflow + stack pattern skills
3	Project Planning	Multi-feature management (optional)

---

The Problem

AI coding assistants have a memory problem:

• No Memory — Every conversation starts from zero. You explain the same things repeatedly.
• Generic Code — Without project context, AI generates boilerplate that doesn't fit your style.
• Contradictory Advice — Unaware of past decisions, AI suggests conflicting patterns.

---

The Solution

Documentation as AI memory — encode your patterns once, every session follows them.

Without This Kit	With This Kit
Re-explain patterns every session	AI reads context automatically
Generic suggestions	Suggestions match your stack
Conflicting recommendations	Decisions preserved in ADRs
No structured workflow	Three workflow tracks

---

AI Agent Workflow (Default)

For feature work, agents should follow this order:

1. Read AGENTS.md for project constraints and routing.
2. Read current feature docs in specs/<feature>/ (spec.md + tasks.md minimum).
3. Load only relevant skills from .agents/skills/.
4. Pull targeted reference docs from docs/ when implementation needs them.
5. Implement, then update reference docs and add an ADR for significant decisions.

---

Quick Start

Solo side project or new to ACDL? Quick Start -> covers the minimum useful setup in three steps.

Fastest Path

npx @acdl/cli init

Then tell your AI assistant:

Bootstrap AGENTS.md for this project.
load skill `acdl`

This installs methodology skills/templates directly into your project. The acdl skill scans your repo, generates AGENTS.md, and creates only the docs/ files your project actually needs. Nothing more.

See Quick Start for the full progressive path, manual setup, and how to add feature workflows when you need them.

Get Started{ .md-button .md-button--primary } View Modules{ .md-button }

---

Three Workflow Tracks

Module 2 (Dev Workflow) provides structured workflows based on what you're building:

Quick Flow

For: Bug fixes, typos, config changes

No templates needed. Just fix and commit.

Standard Flow

For: Any feature that is not a bug fix

Feature spec + task tracking.

Complex Flow

For: New systems, external integrations, or research-heavy work

Research, spec, tasks, ADR.

---

Tool Agnostic

Works with every AI coding agent:

Agent	AGENTS.md	Skills	@docs/ refs
Cursor	Auto-reads	@skill-name	@docs/file.md
Claude Code	CLAUDE.md symlink	/skill-name	Direct read
GitHub Copilot	Auto-reads	Auto-discovered	Direct read
Cline	Via .clinerules	Auto-discovered	Direct read
OpenCode	Auto-reads	Via skill tool	Direct read
Windsurf	Auto-reads	Via UI	Direct read
Aider	Via /read	Not supported	Via /read

Zero tool-specific files required. Optional bridges available (e.g., .mdc rules that point to skills).

---

View All Modules{ .md-button }