{
  "schema_version": "1.1",
  "id": "archive:https://arxiv.org/abs/2611.17316v1",
  "slug": "2611-17316v1-0cco6aj",
  "url": "https://feed7.dev/p/2611-17316v1-0cco6aj",
  "title": "CoReMem: Cost-Internalizing Memory for Successive Coding Tasks with Prospective Valuation",
  "why_included": "CoReMem tests repository memory by measuring whether stored task traces help on the next task. On ChainSWE, it improved a 27B actor’s resolve rate while reducing steps and token use.",
  "summary": "CoReMem selects compact repository memory by comparing task runs with and without trajectory segments, then adapts the agent based on measured memory value. Its memory centers on **repository locations** and **executable checks**.",
  "practical_implication": "For builders running agents across successive tasks, this suggests keeping memory tied to actionable anchors and evaluating whether those anchors reduce work on later tasks.",
  "agent_context": "CoReMem selects compact repository memory by comparing task runs with and without trajectory segments, then adapts the agent based on measured memory value. Its memory centers on **repository locations** and **executable checks**.\n\nFor builders running agents across successive tasks, this suggests keeping memory tied to actionable anchors and evaluating whether those anchors reduce work on later tasks.\n\nOn ChainSWE, the **27B actor** improved resolve rate from **21.0% to 25.7%**, with 17% fewer steps and 11% fewer cache-aware tokens. The result is from one benchmark; the abstract does not establish how well it transfers to other repositories or workflows.",
  "source": {
    "name": "arXiv",
    "url": "https://arxiv.org/abs/2611.17316v1",
    "published_at": "2026-11-12T18:58:22.000Z"
  },
  "source_class": "blog_post",
  "content_type": "Paper",
  "layer": "agent",
  "domains": [
    "coding"
  ],
  "topics": [
    "agent-memory",
    "harness-engineering"
  ],
  "verification": {
    "status": "needs_review",
    "label": "Needs Review",
    "method": "unverified",
    "verified_at": null
  },
  "uncertainty": [
    "On ChainSWE, the **27B actor** improved resolve rate from **21.0% to 25.7%**, with 17% fewer steps and 11% fewer cache-aware tokens. The result is from one benchmark; the abstract does not establish how well it transfers to other repositories or workflows."
  ],
  "connected_context": {
    "meaning": "CoReMem treats repository memory as a measured investment: it tests trajectory segments against later task outcomes and retains actionable locations and checks. Its ChainSWE result suggests this can help across successive coding tasks, but transfer beyond that benchmark remains open.",
    "corpus_size": 1089,
    "generated_at": "2026-11-14T10:54:06.865Z",
    "connections": [
      {
        "title": "Procedural Graphs: Self-Evolving Execution Structures for LLM Agents",
        "source_name": "arXiv",
        "source_url": "https://arxiv.org/abs/2609.09153v1",
        "feed7_url": "https://feed7.dev/p/2609-09153v1-0ee5zbi",
        "reason": "Both make agent memory actionable for later work. Procedural Graphs refine executable workflows from outcomes, while CoReMem evaluates whether specific trajectory segments help subsequent tasks, adding prospective value measurement to the workflow-memory idea."
      },
      {
        "title": "Effective harnesses for long-running agents",
        "source_name": "Anthropic",
        "source_url": "https://www.anthropic.com/engineering/effective-harnesses-for-long-running-agents",
        "feed7_url": "https://feed7.dev/p/effective-harnesses-for-long-running-agents-0xzfs05",
        "reason": "The harness pattern stores progress across sessions and verifies features end to end; CoReMem adds evidence that repository locations and executable checks can be selected by their measured effect on later tasks."
      }
    ]
  },
  "lifecycle": "Current",
  "published_at": "2026-11-12T18:58:22.000Z",
  "modified_at": "2026-11-12T18:58:22.000Z",
  "supersedes": [],
  "expires_at": null,
  "formats": {
    "html": "https://feed7.dev/p/2611-17316v1-0cco6aj",
    "json": "https://feed7.dev/p/2611-17316v1-0cco6aj.json",
    "markdown": "https://feed7.dev/p/2611-17316v1-0cco6aj.md"
  }
}