← All concepts

claudemd pattern

10 articles · 15 co-occurring · 0 contradictions · 0 briefs

Documentation explicitly describes CLAUDE.md auto-discovery in file directory and parent directories as context injection mechanism

Documentation explicitly describes CLAUDE.md auto-discovery in file directory and parent directories as context injection mechanism

Article explicitly uses CLAUDE.md as session persistence mechanism—storing 'project structure, preferences, and past decisions' to prevent interaction reset

Provides concrete CLAUDE.md implementation with domain-specific conventions for Paddle API, showing how system prompt architecture encodes context

Explicitly mentions CLAUDE.md and warns against stale instructions; core example of system prompt persistence

Author explicitly references building a CLAUDE.md context file in previous work and describes it as core to their system. This is a direct instantiation of context engineering best practice.

Article explicitly discusses CLAUDE.md as persistent context layer that survives session resets

Paper explicitly mentions CLAUDE.md and AGENTS.md as vendor-recommended patterns for maintaining context in version control

Article title explicitly mentions 'CLAUDE.md project memory' as tip #1 or major element. This is a direct application of the CLAUDE.md context engineering pattern.

Each agent receives a CLAUDE.md with task specification—direct application of per-agent context engineering

SKILL.md + references/ structure parallels CLAUDE.md approach of encoding system context, but scales it to portable, composable instruction sets. Each skill ships its own metadata plus supporting mate

query this concept
$ db.articles("claudemd-pattern")
$ db.cooccurrence("claudemd-pattern")
$ db.contradictions("claudemd-pattern")