Agent Memory
Manage agent memory across sessions — store, retrieve, and persist context, decisions, and learned patterns.
Use case
Use this skill when:
- The agent needs to remember things between sessions
- Building a persistent knowledge base
- Storing decisions, patterns, and learned behaviors
- Resuming work without losing context
Memory types
- Session memory: What happened in this conversation
- Project memory: Architecture decisions, domain vocabulary, conventions
- User memory: Preferences, patterns, repeated instructions
- Error memory: Past bugs and their solutions
Process
- Capture — Identify what's worth remembering.
- Store — Write to the appropriate memory file (CONTEXT.md, .memory/, etc.).
- Index — Tag with categories for easy retrieval.
- Retrieve — Search memory before starting work.
- Update — Prune stale entries, merge duplicates.
Rules
- Store facts, not fluff.
- Tag every entry with a category and timestamp.
- Prune entries older than the relevance window.
- Never store secrets or credentials in memory files.
- Prefer structured formats (YAML, JSON) over prose for machine-readable memory.
File structure
.memory/
context.md # Project context and decisions
patterns.md # Learned patterns and conventions
errors.md # Past bugs and solutions
user.md # User preferences and habits
Credit
This skill is based on the work of Matt Pocock (@mattpocock). Original source: mattpocock/skills. Licensed under MIT.
How to use?
1. Download
Go to the AGENT KD SKILL gallery and find this skill card. Click Download to save the SKILL.md file to your computer.
2. Install
Copy the downloaded SKILL.md into your agent's skills directory:
| Agent | Path |
|---|---|
| opencode | ~/.config/opencode/skills//SKILL.md |
| Claude Code | ~/.claude/skills//SKILL.md |
| Codex CLI | ~/.codex/skills//SKILL.md |
| Cursor | ~/.cursor/skills//SKILL.md |
| Gemini CLI | ~/.gemini/skills//SKILL.md |
Example for opencode:
ash mkdir -p ~/.config/opencode/skills/<skill-name> cp SKILL.md ~/.config/opencode/skills/<skill-name>/SKILL.md
3. Use
Restart your agent session. The skill will auto-load based on its description. You can also invoke it explicitly by typing @ in your agent.
4. Verify
Ask your agent something related to the skill's purpose. If it responds using the skill's instructions, you're all set.