\r \r

XMemo Memory\r

\r Give your agent durable memory that survives across sessions, projects, and tools. XMemo is a hosted MCP memory service — no local database, no self-hosting required.\r \r

When to use\r

\r

• The task depends on prior decisions, preferences, or project context.\r
• The user asks to remember something for later.\r
• You need to recall conventions, architecture notes, or past fixes before acting.\r
• The user wants TODOs, reminders, or follow-ups tracked across sessions.\r \r

Workflow\r

\r

1. Recall before assuming. Search or recall XMemo context before making decisions that prior memory could inform.\r
2. Save what matters. Store durable facts: decisions, conventions, preferences, architecture notes, action items. Skip transient chat.\r
3. Keep it concise. One clear memory per concept. Prefer structured facts over verbose narratives.\r
4. Confirm before destroying. Always confirm the exact target before delete, forget, or overwrite operations.\r
5. On auth failure, tell the user: "Visit https://xmemo.dev to sign in and get your token, or run xmemo login if the CLI is installed. Set XMEMO_KEY environment variable." Never request raw tokens in chat.\r \r

Available tools\r

\r Core memory operations provided by the XMemo MCP server:\r \r | Tool | Purpose |\r |------|---------|\r | remember | Save a new memory |\r | recall / recall_context | Retrieve relevant memories before answering |\r | search_memory | Search by query |\r | update_memory | Revise existing memory |\r | forget / forget_memory | Delete a memory |\r | redact_memory | Remove sensitive content while keeping audit trail |\r | explain_memory | Show why a memory exists or matched |\r | create_memory_todo | Create a follow-up task |\r | list_memory_todos | List pending TODOs |\r | complete_memory_todo | Mark a TODO done |\r | record_event | Log a milestone or decision |\r | get_timeline | Show recent events |\r | add_expense | Record a ledger entry |\r \r

Good memory candidates\r

\r

• Repository conventions, build/test/deploy commands.\r
• Architecture decisions and their rationale.\r
• Coding style preferences approved by the user.\r
• Release procedures and deployment notes.\r
• TODOs and follow-ups for future sessions.\r
• Bug fix context that might recur.\r \r

Never save\r

\r

• Secrets, tokens, API keys, OAuth codes, cookies.\r
• Private customer data or sensitive PII.\r
• Temporary debugging output.\r
• Large code blocks (link to files instead).\r