{
  "schema": "vedokrok.public-item.v1",
  "release_id": "MHC-RPUB-20260920-75ad787a",
  "url": "/knowledge/evaluate-what-memory-retrieves-not-just-what-it-stores",
  "id": "MHC-D-RESEARCH-0705",
  "version": "0.1.0",
  "title": "Evaluate what memory retrieves, not just what it stores",
  "summary": "A perfect archive can still produce a bad working memory.",
  "kind": "protocol",
  "body": "Create recall tests from real tasks: which stored items should appear, which should stay out, and how high the important item should rank. Measure misses and distracting over-recall separately. When a task fails, inspect whether the needed fact was absent from storage, retrieved poorly or ignored after retrieval; those are different failures.",
  "limits": [
    "Recall targets can overfit known questions; periodically add new tasks and audit whether the retrieval policy generalizes."
  ],
  "topics": [
    "union-ai-context-memory-skills"
  ],
  "intents": [],
  "source_ids": [
    "RS-25CBF175E2E4C6CF"
  ],
  "evidence": [
    {
      "claim": "The same memory-harness work argues that retrieval ranking and recall policy can determine whether stored memory helps or wastes context.",
      "source_id": "RS-25CBF175E2E4C6CF",
      "role": "supports",
      "note": "Results come from a particular research-agent experiment and should be re-tested locally.",
      "locator": "Bad memory spends tokens in the wrong direction; recall policy as evaluation target"
    }
  ],
  "use_when": [
    "A memory-enabled agent still misses old findings or fills context with irrelevant notes."
  ],
  "avoid_when": [
    "Recall targets can overfit known questions; periodically add new tasks and audit whether the retrieval policy generalizes."
  ],
  "example": "For a recurring customer migration, test whether the agent retrieves the current mapping exception without also loading months of unrelated status notes.",
  "check": "Memory quality has a task-level recall and noise test, not only a count of saved items.",
  "steps": [
    "Memory quality has a task-level recall and noise test, not only a count of saved items."
  ],
  "sources": [
    {
      "id": "RS-25CBF175E2E4C6CF",
      "title": "Memory Harnesses for Long-Running Research Agents",
      "url": "https://ai.engineer/talks/R3-anFK1YM8-memory-harnesses-long-running-research-agents"
    }
  ],
  "relations": [],
  "collections": [
    {
      "id": "RC-74B3DD1F3F9BDCB4",
      "title": "Give AI the right context, memory and reusable skills",
      "url": "/collections/give-ai-the-right-context-memory-and-reusable-skills"
    }
  ]
}