agent-memory
Save and recall knowledge across conversations
Claude forgets important details between sessions. This skill provides a persistent memory system for storing discoveries, patterns, and solutions you need to access later. Your AI assistant maintains context that would otherwise be lost.
Download the skill ZIP
Upload in Claude
Go to Settings → Capabilities → Skills → Upload skill
Toggle on and start using
Test it
Using "agent-memory". Remember that this project uses pnpm workspaces and has a husky pre-commit hook that runs lint-staged
Expected outcome:
- ✓ Saved to: .claude/skills/agent-memory/memories/project-context/pmn-workspaces-husky.md
- ✓ Tags added: workspace, husky, pre-commit
- ✓ Summary: Project uses pnpm workspaces with husky pre-commit hook running lint-staged
Using "agent-memory". What do I know about file processing in this project?
Expected outcome:
- ✓ Found 2 relevant memories:
- - large-file-memory-issue.md (tags: performance, worker, memory-leak)
- - iconv-esm-problem.md (tags: dependencies, esm, encoding)
Using "agent-memory". The Node version changed from 18 to 20 - update the deployment memory
Expected outcome:
- ✓ Updated: .claude/skills/agent-memory/memories/project-context/deployment.md
- ✓ Added 'updated: 2026-01-17' to frontmatter
- ✓ Summary: Updated Node version requirement from 18 to 20
Security Audit
SafeThis is a pure documentation skill with no code execution capabilities. It defines memory storage structure and usage guidelines. All 37 static findings are false positives: the scanner incorrectly flagged markdown formatting and YAML field names as security issues. No scripts, network calls, or external commands exist.
Risk Factors
🌐 Network access (1)
📁 Filesystem access (2)
Quality Score
What You Can Build
Remember technical decisions
Document why you chose a specific architecture, workarounds for library bugs, and lessons learned during complex implementations.
Preserve research findings
Save key sources, data patterns, and conclusions from investigations so you can build on previous work in new sessions.
Track infrastructure patterns
Remember deployment configurations, environment-specific quirks, and solutions to recurring operational issues.
Try These Prompts
Remember this for future sessions: [describe what you discovered, why it matters, and when it applies]
What do I know about [topic] from previous conversations? Check my memories.
Document this pattern: [describe the issue, root cause, and solution]. Create a memory with tags for future reference.
Review my memories about [topic]. Consolidate related notes and suggest a better organization structure.
Best Practices
- Include specific context like file paths, version numbers, and error messages in memories
- Use descriptive tags that match how you will search for information later
- Update memories when the information changes or becomes outdated
Avoid
- Saving every trivial detail instead of genuinely useful information
- Using vague summaries that do not help identify relevant memories
- Leaving outdated information that could mislead future sessions