memory-inspect

Auditing agent memory stores involves scanning the .claude/agent-memory directory to compute per-agent stats, list recent entries, detect stale or orphaned files, and generate actionable pruning recommendations. This Skill automates that entire discovery and reporting process.

Cleaning up sr-* agent memory directories requires computing per-agent file counts, total sizes, and last modified dates, then optionally pruning stale files. This Skill provides structured output and recommended actions to maintain memory hygiene across pipelines.

Agent memory hygiene is the practice of inspecting and managing memory stores to remove stale or orphaned files that accumulate over time. You need to prune memory files when agent memory directories grow unbounded or pipeline performance degrades.

Yes, you can check agent memory usage across multiple sr-* agents. The Skill discovers all agent memories under .claude/agent-memory, computes per-agent stats, and displays the five most recent entries per agent to surface usage patterns.

Yes, memory-inspect works without external dependencies or components. It operates directly on the .claude/agent-memory directory structure, enforcing runtime parsing rules and discovery mechanisms internally to generate cleanup guidance.

Automated memory pruning is limited to detecting stale and orphaned files based on parsing rules and modification dates. The Skill provides pruning guidance and recommended actions, but users must evaluate structured output before executing irreversible file deletions.