{
  "markdown": "# MemoryOracle\n\n**Persistent long-term memory for AI agents. Agents never forget.**\n\n10 tools. SQLite FTS5 full-text search. Per-namespace isolation. GDPR-compliant forget. Cross-references with FeedOracle Trust Layer.\n\n```\nnpx -y mcp-remote https://mcp.feedoracle.io/memory/mcp/\n```\n\n## The problem\n\nEvery MCP tool call is stateless. When Claude Desktop restarts, when Cursor closes, when the session ends — the agent forgets everything. No user preferences, no project history, no learned rules, no portfolio tracking.\n\nMemoryOracle gives agents a persistent, searchable memory that survives restarts, works across clients, and gets more valuable over time.\n\n## Tools\n\n| Tool | What it does |\n|------|-------------|\n| `store_memory` | Store a fact, preference, decision, or rule. Deduplicates automatically. |\n| `query_memory` | Full-text search with BM25 ranking and recency boost |\n| `update_fact` | Update a memory with new info. Preserves version history (last 10 changes) |\n| `forget` | Permanently delete by ID, category, age, or everything. GDPR Art. 17 |\n| `summarize_session` | End-of-session capture: summary + extracted facts, decisions, action items |\n| `list_memories` | Browse with filters: category, importance, recency, access frequency |\n| `cross_reference` | Verify a memory against FeedOracle Trust Layer evidence |\n| `memory_stats` | Dashboard: count, storage, categories, most accessed, recent |\n| `export_memories` | Export all as JSON. GDPR Art. 20 data portability |\n| `health_check` | Service status |\n\n## Memory categories\n\n| Category | When to use |\n|----------|------------|\n| `fact` | Verified information: \"USDC is MiCA authorized\" |\n| `preference` | User/agent preferences: \"Conservative allocation, no USDT\" |\n| `decision` | Decisions made: \"Blocked USDT for EU clients\" |\n| `observation` | Patterns noticed: \"RLUSD peg stable for 30 days\" |\n| `rule` | Business rules: \"Never recommend assets with peg deviation > 2%\" |\n| `session` | Session summaries with linked facts |\n| `portfolio` | Portfolio positions and allocations |\n| `action_item` | Open tasks: \"Check RLUSD status in April\" |\n\n## Example: Compliance agent workflow\n\n**Session 1** — Agent does MiCA compliance checks:\n```json\n{\"name\": \"store_memory\", \"arguments\": {\n  \"content\": \"USDT blocked for EU — MiCA NOT AUTHORIZED. Preflight: STOP.\",\n  \"category\": \"decision\", \"importance\": 10,\n  \"tags\": [\"usdt\", \"mica\", \"blocked\"],\n  \"namespace\": \"compliance-bot\"\n}}\n```\n\n**Session 2** — Next day, agent remembers:\n```json\n{\"name\": \"query_memory\", \"arguments\": {\n  \"query\": \"USDT EU compliance status\",\n  \"namespace\": \"compliance-bot\"\n}}\n```\n→ Instantly recalls the decision from yesterday. No re-checking needed.\n\n**End of session** — Auto-capture:\n```json\n{\"name\": \"summarize_session\", \"arguments\": {\n  \"summary\": \"Reviewed stablecoin portfolio. USDT blocked. RLUSD pending.\",\n  \"key_facts\": [\"USDC authorized\", \"EURC authorized\", \"USDT blocked\"],\n  \"decisions\": [\"Block USDT for EU\"],\n  \"action_items\": [\"Check RLUSD in April 2026\"],\n  \"namespace\": \"compliance-bot\"\n}}\n```\n\n## Why this is different\n\n| Feature | MemoryOracle | Others |\n|---------|-------------|--------|\n| **Search** | FTS5 BM25 ranking + recency boost | Basic keyword |\n| **Categories** | 8 structured types | Flat key-value |\n| **Session capture** | Auto-extracts facts, decisions, actions | Manual only |\n| **Trust Layer** | Cross-reference with signed evidence | None |\n| **GDPR** | Art. 17 forget + Art. 20 export | Usually missing |\n| **Namespaces** | Per-agent isolation | Shared global |\n| **History** | Version tracking (last 10 changes) | Overwrite |\n| **TTL** | Auto-expiry with configurable hours | None |\n\n## x402 Pay-per-call\n\n```\nPOST https://tooloracle.io/x402/memory/mcp/\n```\n\n| Tool | Units | Price |\n|------|-------|-------|\n| `store_memory` | 2 | $0.02 |\n| `query_memory` | 5 | $0.05 |\n| `summarize_session` | 8 | $0.08 |\n| `cross_reference` | 5 | $0.05 |\n| `list_memories` | 3 | $0.03 |\n| `update_fact` | 2 | $0.02 |\n| `forget` | 1 | $0.01 |\n\n## Connect\n\n```json\n{\n  \"mcpServers\": {\n    \"memory\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"mcp-remote\", \"https://mcp.feedoracle.io/memory/mcp/\"]\n    }\n  }\n}\n```\n\n## Part of ToolOracle\n\nMemoryOracle is part of [ToolOracle](https://tooloracle.io) — 50+ MCP servers, 530+ tools. Powered by [FeedOracle](https://feedoracle.io) compliance infrastructure.\n\nThe memory layer that makes Decision Preflight, Trust Layer, ISO 20022, and every other tool exponentially more valuable.\n\n\n## Read More\n\n📝 [We Built an AI Agent That Runs 24/7, Never Forgets, and Checks Its Own Work](https://tooloracle.io/blog/autonomous-ai-agent-memory-scheduler-mcp) — How Memory + Scheduler + Decision Preflight work together as one integrated autonomous agent stack. Real tool outputs, no theory.\n\n## License\n\nMIT\n",
  "bytes": 4821,
  "sha": "bc1d123ea0b7fcbe60949ec899afde3b4b0f006a29e82c2524a0fc6b1fc78f42",
  "repo_slug": "tooloracle/memoryoracle",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_io_tooloracle_memoryoracle_16baa1b5/readme"
}