Agent Memory Skill
Enable persistent memory for AI coding agents via Audaxic's MCP server. Agents automatically search and save memories based on behavioral triggers.
Choose Your Agent
| Agent | File | Location |
|---|---|---|
| Universal Reference | AUDAXIC_MEMORY.md | Project root |
| Claude Code | SKILL.md | .claude/skills/audaxic-memory/ |
| Cursor | audaxic-memory.mdc | .cursor/rules/ |
| Windsurf / Devin | audaxic-memory.md | .devin/rules/ |
| Cline | audaxic-memory.md | .clinerules/ |
| GitHub Copilot | audaxic-memory.instructions.md | .github/instructions/ |
| OpenCode | SKILL.md | .opencode/skills/audaxic-memory/ |
| OpenAI Codex | config.toml | ~/.codex/config.toml |
Quick Start
1. Add Skill File
Click your agent above, copy the file content, and place it in the specified location.
2. Test Connection
npx @modelcontextprotocol/inspector --transport http --server-url https://memory-mcp.audaxic.com/mcp --header "Authorization: Bearer ak_your_key_here"
Behavioral Triggers
When to AUTO-SEARCH
Before acting, search for prior context:
- Making architectural decisions → search for prior decisions
- Adding new dependencies → search for existing preferences
- Fixing bugs → search for known issues
- Starting new sessions → search for user preferences
When to AUTO-SAVE
After learning something new:
- User states a preference → save preference
- Decision with rationale → save decision + reasoning
- Complex bug fixed → save solution pattern
- Conventions learned → save conventions
When NOT to Save
- Temporary state (use variables/cache)
- Debug info
- Secrets, API keys, passwords
MCP Tools Reference
| Tool | Parameters | Description |
|---|---|---|
add_memory | text (required), user_id?, metadata? | Store new memory |
search_memories | query (required), user_id?, limit? | Semantic search |
get_memories | user_id?, limit? | List all memories |
delete_memory | memory_ids[] (required) | Delete by IDs |
delete_all_memories | user_id? | Delete all |
Content Best Practices
- One fact per memory — avoid bundling unrelated information
- Natural language — use sentences, not keywords
- Include reasoning — store decisions, not just outcomes
- Be specific — better specificity = better semantic matching
Memory Lifecycle
Add → Search → Update (delete + re-add) → Delete (when obsolete)
Troubleshooting
MCP connection fails
Test your connection:
npx @modelcontextprotocol/inspector --transport http --server-url https://memory-mcp.audaxic.com/mcp --header "Authorization: Bearer ak_your_key_here"
Skill not showing up
- Verify file is in correct location
- Check frontmatter includes
nameanddescription(for SKILL.md format) - Restart your agent after adding the file
Memories not persisting
- Verify MCP server is connected
- Check API key is valid at memory-dashboard.audaxic.com
- Ensure
user_idandproject_idare consistent across sessions
See Also
- MCP Setup — Connect your agent to the MCP server
- MCP Tools — Detailed tool documentation
- Authentication — API key management