AI
13 packages, each available for 11 model families.
agent-rules
This document defines the behavioral operating rules for Claude during software engineering tasks.
context
This document defines how context should be collected, maintained, expanded, and utilized throughout an engineering session.
hallucination
This document defines the anti-hallucination engineering protocol.
memory
This document defines how memory should be managed during engineering tasks.
multi-agent
This document defines how complex engineering tasks should be decomposed into multiple specialized reasoning roles.
optimization
This document defines the engineering optimization methodology.
planning
This document defines the planning methodology for engineering tasks.
prompt-engineering
This document defines how prompts should be constructed for Claude during software engineering tasks.
reasoning
Define how engineering problems should be analyzed before implementation.
self-review
This document defines the self-review methodology for engineering tasks.
spec-driven
Spec-driven development for agents — propose, spec, tasks, verify, archive before any multi-file edit, so the model builds what was agreed and leaves…
tool-use
This document defines how external tools should be used during engineering tasks.
verification
This document defines the verification methodology for software engineering tasks.