{
  "schema_version": "newruntime-topic-hub-v0.1",
  "type": "topic_hub",
  "slug": "agent-memory",
  "title": "Agent memory - New Runtime",
  "description": "Evidence on agent memory as a write loop: what should persist, who can correct it, and how it changes future work.",
  "answer": [
    "Agent memory is not only retrieval; it is a governed write loop that decides what survives a task.",
    "Useful memory needs correction-time updates, provenance, decay, and behavioral evaluation.",
    "The failure mode is treating memory as a passive store while the actual system behavior drifts."
  ],
  "search_intents": [
    "agent memory systems",
    "LLM memory evaluation",
    "company memory AI agents"
  ],
  "status": "featured",
  "last_updated": "2026-07-21",
  "counts": {
    "total": 13,
    "signals": 9,
    "patterns": 1,
    "posts": 1,
    "atlas": 2,
    "sources": 19
  },
  "routes": {
    "html": "https://newruntime.com/topics/agent-memory/",
    "markdown": "https://newruntime.com/topics/agent-memory.md",
    "json": "https://newruntime.com/topics/agent-memory.json"
  },
  "top_sources": [
    "https://aclanthology.org/2024.acl-long.747",
    "https://careersatdoordash.com/blog/doordash-unified-consumer-memory-for-personalization-at-scale",
    "https://developers.openai.com/cookbook/examples/agents_sdk/building_reliable_agents_memory_compaction",
    "https://github.com/AI-Builder-Club/skills",
    "https://github.com/cocoindex-io/cocoindex",
    "https://github.com/langchain-ai/deepagents",
    "https://github.com/langchain-ai/openwiki",
    "https://github.com/openai/openai-cookbook/blob/main/examples/agents_sdk/building_reliable_agents_memory_compaction.ipynb",
    "https://github.com/OpenDataBox/MemoryData",
    "https://github.com/rohitg00/agentmemory",
    "https://github.com/tinyhumansai/openhuman",
    "https://github.com/topoteretes/cognee"
  ],
  "patterns": [
    {
      "kind": "Pattern",
      "slug": "company-memory-needs-write-loops",
      "title": "Company memory needs write and correction loops",
      "description": "Company memory becomes useful when real work continuously updates a governed knowledge layer, not when an agent merely searches a larger document pile.",
      "date": "2026-07-16",
      "topics": [
        "company-memory",
        "agent-memory",
        "knowledge-systems"
      ],
      "source_count": 5,
      "metric": "high",
      "url": "https://newruntime.com/patterns/company-memory-needs-write-loops/"
    }
  ],
  "field_notes": [
    {
      "kind": "Field Note",
      "slug": "cognee-agent-memory-platform",
      "title": "Cognee Packages Agent Memory as a Self-Hosted Knowledge Graph",
      "description": "Cognee matters less as another memory SDK and more as an attempt to package persistent agent memory through ingestion, graph/vector search, ontology, and self-hosting.",
      "date": "2026-07-21",
      "topics": [
        "agent-memory",
        "knowledge-graph",
        "open-source"
      ],
      "source_count": 1,
      "url": "https://newruntime.com/posts/cognee-agent-memory-platform/"
    }
  ],
  "raw_signals": [
    {
      "kind": "Raw Signal",
      "slug": "agent-memory-evaluated-as-system-behavior",
      "title": "Agent memory is being evaluated as system behavior",
      "description": "MemoryData evaluates what an agent stores, retrieves, updates, and forgets across a sequence instead of grading one final answer.",
      "date": "2026-07-06",
      "topics": [
        "agent-memory",
        "evals",
        "knowledge-systems"
      ],
      "source_count": 2,
      "metric": "structural",
      "url": "https://newruntime.com/signals/agent-memory-evaluated-as-system-behavior/"
    },
    {
      "kind": "Raw Signal",
      "slug": "codebase-memory-combines-search-with-a-repository-map",
      "title": "Codebase Memory Combines Search with a Repository Map",
      "description": "A code retrieval tool adds structural repository context to text search so an agent can locate relevant code without loading the whole project into context.",
      "date": "2026-07-03",
      "topics": [
        "code-search",
        "agent-memory",
        "context-engineering"
      ],
      "source_count": 2,
      "metric": "notable",
      "url": "https://newruntime.com/signals/codebase-memory-combines-search-with-a-repository-map/"
    },
    {
      "kind": "Raw Signal",
      "slug": "agent-memory-needs-its-own-review-loop",
      "title": "Agent Memory Needs Its Own Review Loop",
      "description": "Persistent memory can amplify mistakes across future tasks, so edits, provenance, and correction paths need explicit operator control.",
      "date": "2026-07-02",
      "topics": [
        "agent-memory",
        "review",
        "safety"
      ],
      "source_count": 1,
      "metric": "notable",
      "url": "https://newruntime.com/signals/agent-memory-needs-its-own-review-loop/"
    },
    {
      "kind": "Raw Signal",
      "slug": "deep-agents-plan-files-memory-subagents",
      "title": "Deep Agents combines plans, files, memory, and subagents",
      "description": "The open-source Deep Agents stack packages common long-horizon primitives on top of LangGraph and LangChain.",
      "date": "2026-07-01",
      "topics": [
        "harness-engineering",
        "orchestration",
        "agent-memory"
      ],
      "source_count": 1,
      "metric": "notable",
      "url": "https://newruntime.com/signals/deep-agents-plan-files-memory-subagents/"
    },
    {
      "kind": "Raw Signal",
      "slug": "doordash-memory-is-versioned-personalization-layer",
      "title": "DoorDash makes memory a versioned personalization layer",
      "description": "DoorDash builds structured memory blocks from behavior and stores model, schema, timestamp, prompt, and response lineage for each component.",
      "date": "2026-06-24",
      "topics": [
        "agent-memory",
        "personalization",
        "data-lineage"
      ],
      "source_count": 1,
      "metric": "structural",
      "url": "https://newruntime.com/signals/doordash-memory-is-versioned-personalization-layer/"
    },
    {
      "kind": "Raw Signal",
      "slug": "openhuman-builds-a-personal-harness-around-local-memory",
      "title": "OpenHuman Builds a Personal Harness Around Local Memory",
      "description": "OpenHuman combines personal data, memory, tools, and an operator layer into a durable system instead of another isolated assistant chat.",
      "date": "2026-05-24",
      "topics": [
        "personal-agents",
        "local-first",
        "agent-memory"
      ],
      "source_count": 2,
      "metric": "notable",
      "url": "https://newruntime.com/signals/openhuman-builds-a-personal-harness-around-local-memory/"
    },
    {
      "kind": "Raw Signal",
      "slug": "agentmemory-shares-state-across-codex-claude-code-and-cursor",
      "title": "AgentMemory Shares State Across Codex, Claude Code, and Cursor",
      "description": "A common memory layer reduces the cost of switching coding clients and moves context ownership outside any one model interface.",
      "date": "2026-05-19",
      "topics": [
        "agent-memory",
        "coding-agents",
        "interoperability"
      ],
      "source_count": 1,
      "metric": "notable",
      "url": "https://newruntime.com/signals/agentmemory-shares-state-across-codex-claude-code-and-cursor/"
    },
    {
      "kind": "Raw Signal",
      "slug": "openai-shows-memory-compaction-for-long-running-agents",
      "title": "OpenAI Shows Memory Compaction for Long-Running Agents",
      "description": "The Agents SDK pattern treats compaction and state transfer as explicit reliability mechanisms for work that outlives one context window.",
      "date": "2026-05-07",
      "topics": [
        "agent-memory",
        "compaction",
        "long-running-agents"
      ],
      "source_count": 2,
      "metric": "notable",
      "url": "https://newruntime.com/signals/openai-shows-memory-compaction-for-long-running-agents/"
    },
    {
      "kind": "Raw Signal",
      "slug": "cocoindex-keeps-agent-context-fresh-without-full-reindexing",
      "title": "CocoIndex Keeps Agent Context Fresh Without Full Reindexing",
      "description": "CocoIndex turns source changes into incremental dataflow updates for live agent context and code-aware retrieval.",
      "date": "2026-05-06",
      "topics": [
        "agent-memory",
        "retrieval",
        "data-infrastructure"
      ],
      "source_count": 1,
      "metric": "notable",
      "url": "https://newruntime.com/signals/cocoindex-keeps-agent-context-fresh-without-full-reindexing/"
    }
  ],
  "atlas_records": [
    {
      "kind": "Shift",
      "slug": "shared-context-infrastructure",
      "title": "Client-local memory -> shared context infrastructure",
      "description": "Context is shifting from private memory inside one AI client into a governed layer between agents, IDEs, documents, and tools.",
      "date": "2026-07-21",
      "topics": [
        "context-engineering",
        "agent-memory",
        "mcp"
      ],
      "source_count": 2,
      "metric": "emerging",
      "url": "https://newruntime.com/shifts/shared-context-infrastructure/"
    },
    {
      "kind": "Graveyard",
      "slug": "manual-context-handoff",
      "title": "Manual context handoff between AI clients",
      "description": "The ritual of retelling one AI client what another already understood becomes optional when context moves into a shared layer.",
      "date": "2026-07-21",
      "topics": [
        "context-engineering",
        "agent-memory"
      ],
      "source_count": 1,
      "metric": "challenged",
      "url": "https://newruntime.com/graveyard/manual-context-handoff/"
    }
  ]
}