speckit-memorylint-run

To separate architecture rules from infrastructure rules in AGENTS.md, you can use a memory boundary lint workflow to extract architectural constraints, prune them from the file, and enrich the remaining infrastructure sections with operational conventions.

AI agent memory boundary hygiene in spec-kit projects is the practice of preventing architectural drift by cleanly separating architectural constraints from infrastructure rules within AGENTS.md and .specify/memory/constitution.md to maintain consistent agent behavior.

You can prune out-of-scope architectural constraints from AGENTS.md by running an audit that identifies architecture-layer rules like state management and API design, removes them from the file, and adds a pointer directing agents to constitution.md.

You can enrich AGENTS.md with missing build and test workflow conventions by auditing the current workspace context, identifying absent operational essentials, and injecting infrastructure guidance covering environment setup, lint, and Git workflows.

Memory governance linting in spec-kit based projects typically requires a .specify/memory/constitution.md file, because the workflow extracts out-of-scope architectural rules from AGENTS.md and adds a pointer to this constitution file for proper relocation.

An AI agent may exhibit architectural drift when AGENTS.md mixes infrastructure rules with architecture constraints, which a memory boundary audit resolves by extracting layering and state management rules to maintain clean, focused agent memory boundaries.