Use when Abu wants Claude Code to create, audit, or update user-level ~/.claude/CLAUDE.md, repo CLAUDE.md, .claude/CLAUDE.md, CLAUDE.local.md, or .claude/rules/*.md files. Keeps instructions small, practical, and sourced from project evidence, while routing detailed workflows to skills instead of stuffing long process manuals into always-on memory.
How this skill is triggered — by the user, by Claude, or both
Slash command
/abu-context-engineering:claude-md-authorThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Use this skill to create or maintain Claude Code instruction files.
Use this skill to create or maintain Claude Code instruction files.
CLAUDE.md is always-on project memory, not the whole operating system. Put durable facts, commands, conventions, and short behavior rules there. Put multi-step procedures in skills, deterministic enforcement in hooks or CI, and project artifacts in <repo>/.thoughts/.
~/.claude/CLAUDE.md.~/.claude/rules/*.md.<repo>/CLAUDE.md or <repo>/.claude/CLAUDE.md.<repo>/.claude/rules/*.md.<repo>/CLAUDE.local.md.Claude Code reads these files as context, not as hard enforcement. Use hooks or CI for rules that must always run.
User-level ~/.claude/CLAUDE.md:
Repo CLAUDE.md or .claude/CLAUDE.md:
.thoughts/ artifacts instead of copying them..claude/rules/*.md:
CLAUDE.md noisy.Do not put these in Claude instruction files:
CLAUDE.md versus skills, hooks, CI, rules, or .thoughts.# CLAUDE.md
## Project Snapshot
## Working Rules
## Commands
## Quality Gates
## Project Context
## Skill Routing
Keep the file short enough to stay useful in every session. If artifact boundaries are unclear, read ../../references/operating-model.md from the plugin root.
npx claudepluginhub blockchain-oracle/abu-context-engineering --plugin abu-context-engineeringCreates bite-sized, testable implementation plans from specs or requirements, with file structure and task decomposition. Activates before coding multi-step tasks.