{
  "schema_version": 1,
  "artifact": "TVIJO-EVAL-COMMS-RADAR-20261109 New Runtime fixed-sample verification",
  "decided_at": "2026-11-09T04:36:00Z",
  "summary": {
    "classification_counts_exactly_over_selected_10": {
      "confirmed": 4,
      "partially_confirmed": 3,
      "stale": 2,
      "unverified": 1,
      "duplicate": 0,
      "sum": 10
    },
    "records_with_supporting_primary_source": 9,
    "records_with_official_primary_source_reviewed_but_no_support_for_main_claim": 1,
    "duplicate_flags": 0,
    "records_with_forecast_claims": 0,
    "utility_criteria": [
      {
        "criterion_id": "NR-USE-1",
        "criterion": "At least 8 of the 10 fixed-sample core claims are confirmed by primary evidence and are not stale.",
        "result": "not_met",
        "evidence": "4/10 are classified confirmed; 3 are partial, 2 stale, and 1 unverified.",
        "reason": "Partial, stale, and unverified items do not count toward the threshold."
      },
      {
        "criterion_id": "NR-USE-2",
        "criterion": "Every interpretation and inference in this audit is explicitly labeled separately from reported facts.",
        "result": "met",
        "evidence": "All 10 records use separate reported_facts, publisher_interpretation, analyst_inference, and forecast_claims fields.",
        "reason": "The audit preserves New Runtime's Field Note/Analysis distinction while also marking the auditor's own applicability judgments."
      },
      {
        "criterion_id": "NR-USE-3",
        "criterion": "Source-selection counts are exact, duplicate flags are separate, and no substitutes are added.",
        "result": "met",
        "evidence": "The 10 selected URLs and RSS publication timestamps match the frozen selection file; duplicate_flags count is 0.",
        "reason": "Only the frozen selected records are assessed."
      }
    ],
    "promotion_readiness": "not_met",
    "promotion_decision": "NOT_PROMOTED: threshold not met",
    "recommendation_to_sol": "Treat New Runtime as a discovery lead, not a verified technical source. Do not add it to the existing radar registry on this sample alone; reassess only with a larger fresh sample and primary-source checks.",
    "limitations": [
      "This is one fixed 10-item snapshot, not a longitudinal accuracy or freshness study.",
      "First-party product pages and authored case studies verify that claims were published by their owners; they do not independently establish real-world performance or user counts.",
      "No New Runtime instructions or suggested commands were executed.",
      "No registry, crawler, production system, or recurring job was changed."
    ]
  },
  "records": [
    {
      "selection_index": 1,
      "claim_id": "NR-20261109-01",
      "title": "vork is a headless Rust/Slang 3D engine where coding agents write scenes as code and render on GPU.",
      "newruntime_url": "https://newruntime.com/posts/vork-is-a-headless-rust-slang-3d-engine-12ecec/",
      "publisher_type": "field_note",
      "rss_published_at_utc": "2026-11-09T03:00:02Z",
      "primary_event_date": {
        "value": "2026-11-07T20:41:52Z",
        "basis": "GitHub commit a18cac5 is titled 'vork 0.2.0'; this is the nearest pinned project event before the RSS publication."
      },
      "primary_sources": [
        {
          "url": "https://github.com/voyager-so/vork",
          "kind": "official_project_repository",
          "version": "0.2.0 at commit a18cac5d68eae4b6d8d7c0126c44722b02edb897",
          "source_date": "2026-11-07T20:41:52Z",
          "capture": "vork_v0.2.0_readme.raw",
          "capture_sha256": "ebdf2362e9b3c774c231f011f0b67a87b17bc4db788cfe6ca429dce341692065"
        },
        {
          "url": "https://api.github.com/repos/voyager-so/vork/commits?per_page=5",
          "kind": "official_version_history",
          "capture": "vork_main_commits.json",
          "capture_sha256": "593f6a69abfa535071355583a3842ed6c5f1e12fa4cda87e36f7b967b51a2d4f"
        }
      ],
      "reported_facts": [
        "The project describes itself as a headless 3D/VFX engine for agents, implemented in Rust and Slang, with GPU rendering.",
        "The commit current around the story date is v0.2.0; v0.3.0 landed at 2026-11-09T03:31:50Z, after this RSS item's 03:00:02Z publication."
      ],
      "publisher_interpretation": [
        "New Runtime frames the tool as code-authored scene creation rather than driving a conventional GUI."
      ],
      "analyst_inference": [
        {
          "text": "Low direct fit for Tvijo's coding-agent communication/evaluation needs; it may be relevant only if the product later adds agent-authored 3D/VFX workflows.",
          "label": "inference"
        }
      ],
      "forecast_claims": [],
      "result": "confirmed",
      "limitations": [
        "Repository description and examples establish what the project claims and contains, not quality or reliability of rendered output.",
        "Current main has advanced beyond the 0.2.0 snapshot; the 0.3.0 commit postdates the RSS time."
      ],
      "contradictions": [],
      "recheck_condition": "Recheck the pinned README and latest release before discussing current capabilities; distinguish features present at publication from later commits.",
      "duplicate_flags": []
    },
    {
      "selection_index": 2,
      "claim_id": "NR-20261109-02",
      "title": "Continuity Research released Tess-4.5-27B, a 27B Apache-2.0 fine-tune of Qwen3.8-27B for agentic coding.",
      "newruntime_url": "https://newruntime.com/posts/continuity-research-released-tess-4-5-27b-a-1c8d16/",
      "publisher_type": "field_note",
      "rss_published_at_utc": "2026-11-09T00:00:02Z",
      "primary_event_date": {
        "value": "2026-10-18T18:32:14Z",
        "basis": "Hugging Face model API createdAt; lastModified 2026-10-18T23:29:35Z."
      },
      "primary_sources": [
        {
          "url": "https://huggingface.co/migtissera/Tess-4.5-27B",
          "kind": "publisher_model_card",
          "version": "revision 91b231133470bd147aac8036ed94d35e9bf1dc13",
          "source_date": "2026-10-18",
          "capture": "tess_readme.raw",
          "capture_sha256": "f4e2237c13c137034f8602c697e58aeb62bafa4e37bf785a32ca441772de78b9"
        },
        {
          "url": "https://huggingface.co/api/models/migtissera/Tess-4.5-27B",
          "kind": "official_model_metadata",
          "version": "91b231133470bd147aac8036ed94d35e9bf1dc13",
          "capture": "tess_model.json",
          "capture_sha256": "b61566cffb019c646162a0c66dccbcfdd6d97f7dcd42e897cc83df3cc77548b8"
        }
      ],
      "reported_facts": [
        "The model card identifies a 27B model, Apache-2.0 license, Qwen/Qwen3.8-27B base, and coding/agentic tags."
      ],
      "publisher_interpretation": [
        "The model card describes the model as built for engineering and agentic work; its reported BenchLocal scores and speed figures are publisher-authored results, not independently reproduced here."
      ],
      "analyst_inference": [
        {
          "text": "Useful as a candidate to monitor for local-model and harness compatibility, but it is not evidence for Tvijo's four-shell benchmark or a coding-agent quality ranking.",
          "label": "inference"
        }
      ],
      "forecast_claims": [],
      "result": "confirmed",
      "limitations": [
        "No model was installed or run; no benchmark, weight integrity, or inference performance was independently checked."
      ],
      "contradictions": [],
      "recheck_condition": "Recheck the model revision, base-model declaration, license, and benchmark protocol before citing performance or current availability.",
      "duplicate_flags": []
    },
    {
      "selection_index": 3,
      "claim_id": "NR-20261109-03",
      "title": "Zed 1.24.2 shipped Oct 14 with Git tags, Terminal tabs in the Agent Panel and File Finder pre-fill",
      "newruntime_url": "https://newruntime.com/analysis/research-zed-1-24-2-shipped-oct-14-with-git-tags-terminal-tabs-in-the-agent-panel-and/",
      "publisher_type": "analysis",
      "rss_published_at_utc": "2026-11-08T22:00:00Z",
      "primary_event_date": {
        "value": "2026-10-14T16:43:09Z",
        "basis": "GitHub release API published_at for v1.24.2; Zed's stable release page separately gives Oct 14, 2026."
      },
      "primary_sources": [
        {
          "url": "https://github.com/zed-industries/zed/releases/tag/v1.24.2",
          "kind": "official_release",
          "version": "v1.24.2",
          "source_date": "2026-10-14T16:43:09Z",
          "capture": "zed_release.json",
          "capture_sha256": "231c8611d74719d792d3b9ac29eeee51b974c334955233cbe94e1d7ebe8ff0b1"
        },
        {
          "url": "https://zed.dev/releases/stable/1.24.2",
          "kind": "official_release_notes",
          "version": "1.24.2",
          "capture": "zed_stable.raw",
          "capture_sha256": "0596e59ef66a48a7053d4c5fb21758eb560b0aa6b4e733d9cd0d842aea66eed0"
        },
        {
          "url": "https://zed.dev/releases/stable",
          "kind": "current_stable_index",
          "version": "1.27.0 observed",
          "capture": "zed_current_stable.raw",
          "capture_sha256": "e47573db7f51aad1c47ef209843c05846fdecfa10c176af8475bc7ea5c2fcfe3"
        },
        {
          "url": "https://api.github.com/repos/zed-industries/zed/releases/tags/v1.24.4",
          "kind": "later_official_release",
          "version": "v1.24.4",
          "source_date": "2026-10-20T10:34:04Z",
          "capture": "v1.24.4.json",
          "capture_sha256": "cf0c770a21c3a71964b99d41d756c9bb9aee982d09be293b94da0775732e49de"
        },
        {
          "url": "https://api.github.com/repos/zed-industries/zed/releases/tags/v1.27.0",
          "kind": "later_official_release",
          "version": "v1.27.0",
          "source_date": "2026-11-04T17:04:23Z",
          "capture": "v1.27.0.json",
          "capture_sha256": "79c99e53c38f9ce9792af66cae0aaa887270f82077cd76f7deb21d6e82712772"
        }
      ],
      "reported_facts": [
        "The official release date and three named features match the Zed release notes.",
        "Later official releases 1.24.4 (Oct 20) and 1.27.0 (Nov 4) both predate the Nov 8 aggregator item, so 1.24.2 was already superseded when the story appeared."
      ],
      "publisher_interpretation": [
        "New Runtime's analysis says a 1.24.1 link in its reviewed evidence was not a stable release and had caused a later changelog excerpt to be mixed in."
      ],
      "analyst_inference": [
        {
          "text": "The release facts are correct historically, but this item is stale as a current release update and has little direct relevance to Tvijo's internal agent runtime.",
          "label": "inference"
        }
      ],
      "forecast_claims": [],
      "result": "stale",
      "limitations": [
        "Historical release accuracy does not establish that the item was timely when republished on Nov 8.",
        "The v1.24.1 GitHub release API lookup returned 404 and the current Zed route resolves to the stable index; that is not proof that no historical development/pre-release build existed."
      ],
      "contradictions": [
        "The Nov 8 aggregator publication followed later stable versions; current official index already showed 1.27.0 at audit time."
      ],
      "recheck_condition": "For a current radar, verify the latest stable version and compare the release event date with the aggregator date; retain old releases only as explicitly historical notes.",
      "duplicate_flags": []
    },
    {
      "selection_index": 4,
      "claim_id": "NR-20261109-04",
      "title": "AgentAudit publishes MCP server 3.0.0 and SDK 1.2.0 with Merkle-batched agent audit logs on Base",
      "newruntime_url": "https://newruntime.com/posts/agentaudit-publishes-mcp-server-3-0-0-and-3ee038/",
      "publisher_type": "field_note",
      "rss_published_at_utc": "2026-11-08T21:00:02Z",
      "primary_event_date": {
        "value": "2026-11-08T16:36:23.414Z",
        "basis": "NPM registry time for @agentauditai/mcp-server 3.0.0; SDK 1.2.0 was published at 2026-11-08T16:33:05.469Z."
      },
      "primary_sources": [
        {
          "url": "https://registry.npmjs.org/%40agentauditai%2fmcp-server",
          "kind": "official_package_registry",
          "version": "3.0.0",
          "source_date": "2026-11-08T16:36:23.414Z",
          "package_shasum": "1ca2fbb17237aed5227065c3cd092fb2c89e2e1b",
          "capture": "agentaudit_server_full.json",
          "capture_sha256": "8567db7834408f81978ae4401c139eca8aed674d30de4fab25579bf2876baad2"
        },
        {
          "url": "https://registry.npmjs.org/%40agentauditai%2fsdk",
          "kind": "official_package_registry",
          "version": "1.2.0",
          "source_date": "2026-11-08T16:33:05.469Z",
          "package_shasum": "55b6d2376b6bf48647e3d21ae73f4a60af93d532",
          "capture": "agentaudit_sdk_full.json",
          "capture_sha256": "36dc5951d32b6d3760f0264f9246001c3225d000dead730b78d34813eb4d6ae3"
        }
      ],
      "reported_facts": [
        "The registry lists both versions as latest and gives matching Nov 8 publication times.",
        "The package README describes audit_action/get_audit_log/verify_proof tools and Base AuditVault v2; it also states that the vault does not authenticate submitters and that evidence bundles must be retained."
      ],
      "publisher_interpretation": [
        "New Runtime presents this as an MCP-based audit/evidence interoperability pattern, while warning that the address supplied by a caller does not prove the caller's identity."
      ],
      "analyst_inference": [
        {
          "text": "Potentially relevant to audit-evidence design, but an on-chain third-party package is not a substitute for AIOS's authorized journal, lease, tenant scope, or controller receipts.",
          "label": "inference"
        }
      ],
      "forecast_claims": [],
      "result": "confirmed",
      "limitations": [
        "Package metadata and README verify the release and declared behavior, not contract security, a live transaction, identity attribution, or legal compliance.",
        "NPM's read-only package endpoint was used; the package was not installed or executed."
      ],
      "contradictions": [],
      "recheck_condition": "Recheck package version, registry integrity, deployed contract code/access controls, and explicit submitter identity before any technical comparison or adoption.",
      "duplicate_flags": []
    },
    {
      "selection_index": 5,
      "claim_id": "NR-20261109-05",
      "title": "Nous Research released hermes-security, an official Hermes Agent plugin for local repo security review.",
      "newruntime_url": "https://newruntime.com/posts/nous-research-released-hermes-security-an-official-hermes-d55f0e/",
      "publisher_type": "field_note",
      "rss_published_at_utc": "2026-11-08T21:00:02Z",
      "primary_event_date": {
        "value": null,
        "basis": "Nous's current marketplace listing verifies availability but does not provide a stable initial publication timestamp."
      },
      "primary_sources": [
        {
          "url": "https://hermes-agent.nousresearch.com/marketplace/hermes-security",
          "kind": "official_marketplace_listing",
          "version": "0.1.2 at 2026-11-09T04:27Z observation; New Runtime reported 0.1.1 at its publication snapshot",
          "capture": "hermes_security.raw",
          "capture_sha256": "eaa805481b8d5fea4b83906bdced944695a7a56d8283b20ee1b083628e1b7bcf"
        }
      ],
      "reported_facts": [
        "Nous's marketplace labels hermes-security Official and describes parallel subagent reviewers, evidence-backed reports, standard/deep/diff modes, export, and optional reproduction tests.",
        "The listing explicitly says the configured model provider receives the reviewed code and repository test runs are not sandboxed."
      ],
      "publisher_interpretation": [
        "New Runtime calls this a structured review workflow for existing Hermes users."
      ],
      "analyst_inference": [
        {
          "text": "Relevant to security-review workflow research, but the provider data path and test execution scope should be reviewed before any use on Tvijo repositories.",
          "label": "inference"
        }
      ],
      "forecast_claims": [],
      "result": "confirmed",
      "limitations": [
        "This confirms the official listing and stated behavior, not vulnerability-finding quality or safety in a real repository.",
        "The current marketplace version differs from the 0.1.1 version reported by New Runtime, consistent with a possible post-publication update; release history was not independently pinned."
      ],
      "contradictions": [],
      "recheck_condition": "Check the Nous marketplace changelog and exact version before citing the feature set; re-review provider data handling and test isolation.",
      "duplicate_flags": []
    },
    {
      "selection_index": 6,
      "claim_id": "NR-20261109-06",
      "title": "A Habr case study describes a five-layer memory system behind an assistant used by about 1",
      "newruntime_url": "https://newruntime.com/posts/a-habr-case-study-describes-a-five-layer-cc496d/",
      "publisher_type": "field_note",
      "rss_published_at_utc": "2026-11-08T18:00:02Z",
      "primary_event_date": {
        "value": "2026-11-08T11:46:39Z",
        "basis": "Habr JSON-LD datePublished 2026-11-08T14:46:39+03:00."
      },
      "primary_sources": [
        {
          "url": "https://habr.com/ru/articles/1105098/",
          "kind": "first_person_case_study",
          "version": "article 1105098",
          "source_date": "2026-11-08T11:46:39Z",
          "capture": "hab r_article.raw",
          "capture_sha256": "7d6dd6db7315493c720541684220bd39f7120c9b1181229fd9122a61b5826d22"
        }
      ],
      "reported_facts": [
        "The author describes a Telegram/MAX/VK assistant and a multi-layer memory architecture, and self-reports about 1,200 users, 11,674 archived records over 134 subjects, and 16,333 PostgreSQL rounds for 657 users.",
        "The author explicitly says a scheduled production skill-decay run could not be confirmed."
      ],
      "publisher_interpretation": [
        "New Runtime's practical lesson is to preserve original conversations, retrieve bounded context, and keep memory access rules explicit."
      ],
      "analyst_inference": [
        {
          "text": "Useful as a practitioner case study for memory and access-policy design; metrics and production operation remain the author's unverified report.",
          "label": "inference"
        }
      ],
      "forecast_claims": [],
      "result": "partially_confirmed",
      "limitations": [
        "The article is primary to the author's account, not independent evidence of user counts, data volumes, system reliability, or performance.",
        "The author distinguishes implemented layers from ideas still in a library; do not describe all five as production-verified."
      ],
      "contradictions": [],
      "recheck_condition": "Seek a dated first-party artifact or independent evidence before treating scale or production-status metrics as verified; preserve explicit implemented/planned distinctions.",
      "duplicate_flags": []
    },
    {
      "selection_index": 7,
      "claim_id": "NR-20261109-07",
      "title": "WakeVane: open-source, event-driven harness for long-running coding agents, built on Databricks' Omnigent (alpha)",
      "newruntime_url": "https://newruntime.com/posts/wakevane-open-source-event-driven-harness-for-long-c81475/",
      "publisher_type": "field_note",
      "rss_published_at_utc": "2026-11-08T18:00:02Z",
      "primary_event_date": {
        "value": "2026-11-08T14:29:54Z",
        "basis": "Repository dist-branch commit eba874c; repository itself was created 2026-10-22. Official site was captured on Nov 9; it does not expose a stable launch timestamp."
      },
      "primary_sources": [
        {
          "url": "https://www.wakevane.com/",
          "kind": "publisher_product_site",
          "version": "Alpha; no semver stated",
          "capture": "wakevane_home.raw",
          "capture_sha256": "6153f617e7ef43f65c20f8ccf32b190afa209d096b48b51545cf95b253c33013"
        },
        {
          "url": "https://github.com/juanpasaflipz/wakevane",
          "kind": "publisher_repository",
          "version": "dist branch commit eba874c7d25aa1f7bbd3ecb587b88c494b6eb0c6",
          "source_date": "2026-11-08T14:29:54Z",
          "capture": "wakevane_repo.json",
          "capture_sha256": "ba22320a65dbe69040e806b6ed7536b06d1dd7689bd2b8b6378e430400056a69"
        },
        {
          "url": "https://www.databricks.com/blog/introducing-omnigent-meta-harness-combine-control-and-share-your-agents",
          "kind": "upstream_vendor_announcement",
          "version": "Omnigent",
          "capture": "web search result turn21search9",
          "source_date": "2026-06"
        }
      ],
      "reported_facts": [
        "The site and GitHub repository describe WakeVane as an Apache-2.0 alpha harness that responds to events and builds on Omnigent.",
        "The product site says its illustrated sequence is not live activity and that its first public-repository run is not yet available."
      ],
      "publisher_interpretation": [
        "New Runtime interprets WakeVane as an event-triggered Omnigent distribution with spending visibility; it explicitly says no productivity or cost gain has been demonstrated."
      ],
      "analyst_inference": [
        {
          "text": "Conceptually adjacent to long-running agent orchestration, but the site states its public run is unavailable; do not treat product claims as a verified performance result or an AIOS-ready integration.",
          "label": "inference"
        }
      ],
      "forecast_claims": [],
      "result": "partially_confirmed",
      "limitations": [
        "Product description and source repository establish publisher claims, not a successful live event-to-agent run, safety boundaries, or cost reduction.",
        "No code was cloned, installed, or executed."
      ],
      "contradictions": [],
      "recheck_condition": "Require a pinned repository version and independently inspect a reproducible event-triggered run, policy enforcement, sandbox, and cost evidence.",
      "duplicate_flags": []
    },
    {
      "selection_index": 8,
      "claim_id": "NR-20261109-08",
      "title": "OpenAI did shelve GPT-6.1 Astra over alignment failures, but has published no evaluation of it",
      "newruntime_url": "https://newruntime.com/analysis/research-openai-did-shelve-gpt-6-1-astra-over-alignment-failures-but-has-published-no/",
      "publisher_type": "analysis",
      "rss_published_at_utc": "2026-11-08T17:00:00Z",
      "primary_event_date": {
        "value": "2026-09-28 (secondary press-reported only)",
        "basis": "The New Runtime analysis attributes the date to secondary press reports; no official OpenAI primary report confirming a GPT-6.1 Astra cancellation was found."
      },
      "primary_sources": [
        {
          "url": "https://openai.com/index/gpt-6-astra/",
          "kind": "official_model_announcement",
          "version": "GPT-6 Astra; page later references GPT-6.1 Sol",
          "source_date": "2026-09-03 / updated 2026-09-29",
          "support_for_claim": "does_not_confirm_GPT_6_1_Astra_shelving",
          "capture": "openai_astra_launch.raw",
          "capture_error": "HTTP 403 to direct capture; official page checked through web search result turn20search2"
        },
        {
          "url": "https://alignment.openai.com/misalignment-reports/",
          "kind": "official_safety_report_index",
          "version": "GPT-6.1 Sol and GPT-6 Astra reports are listed; no GPT-6.1 Astra report found",
          "capture": "openai_alignment_reports.raw",
          "capture_sha256": "215cfe0522dcd6681da656b2784ebd0bb003b09cf8e36170c2f121b3a90d57d9"
        },
        {
          "url": "https://openai.com/index/safety-overview-gpt-6-astra/",
          "kind": "official_safety_overview",
          "version": "GPT-6 Astra",
          "support_for_claim": "documents_predecessor_only",
          "source_date": "2026-09-03",
          "capture": "web search result turn20search1"
        }
      ],
      "reported_facts": [
        "The New Runtime analysis says secondary outlets reported a decision not to release a planned GPT-6.1 Astra and cites a safety-chief quote; the article itself acknowledges no published evaluation of that named model."
      ],
      "publisher_interpretation": [
        "The analysis treats scope control and accurate reporting as important failure modes for agents with tools."
      ],
      "analyst_inference": [
        {
          "text": "The exact claim cannot be accepted under the requested official-source rule; official OpenAI pages instead identify GPT-6 Astra and GPT-6.1 Sol, a material model-name/version mismatch.",
          "label": "inference"
        }
      ],
      "forecast_claims": [],
      "result": "unverified",
      "limitations": [
        "No official OpenAI source was found confirming GPT-6.1 Astra existed, was shelved, or had the quoted evaluation result.",
        "OpenAI's predecessor materials about GPT-6 Astra do not establish behavior of a supposed GPT-6.1 Astra model; secondary reporting is not substituted as primary evidence."
      ],
      "contradictions": [
        "Official OpenAI material names GPT-6 Astra and GPT-6.1 Sol; the aggregator claim names GPT-6.1 Astra. This is a version mismatch, not an explicit official denial."
      ],
      "recheck_condition": "Recheck only if OpenAI publishes a primary announcement, system card, or signed official statement about GPT-6.1 Astra and its evaluation.",
      "duplicate_flags": []
    },
    {
      "selection_index": 9,
      "claim_id": "NR-20261109-09",
      "title": "SuperARIS v0.1.0 preview adds a persistent multi-agent research workbench on ARIS",
      "newruntime_url": "https://newruntime.com/posts/superaris-v0-1-0-preview-adds-a-persistent-615c33/",
      "publisher_type": "field_note",
      "rss_published_at_utc": "2026-11-08T13:00:03Z",
      "primary_event_date": {
        "value": "2026-11-08T05:12:40Z repository creation; README revision 92118fef dated 09:54:38Z",
        "basis": "GitHub repository metadata and README history. The current 504fe48 README revision is later, 2026-11-08T18:08:07Z."
      },
      "primary_sources": [
        {
          "url": "https://github.com/wanshuiyin/SuperARIS",
          "kind": "publisher_repository",
          "version": "Preview badge v0.1.0; repository created 2026-11-08T05:12:40Z; no GitHub release or tag observed",
          "capture": "superaris_repo.json",
          "capture_sha256": "6386af8eb05ff08272a4dcb30ea2491ba1474258a7e4790b1e403a619050595a"
        },
        {
          "url": "https://raw.githubusercontent.com/wanshuiyin/SuperARIS/main/README.md",
          "kind": "publisher_readme",
          "version": "current audited commit 504fe48892f452a37b027dd5d830e64a001cf32b",
          "source_date": "2026-11-08T18:08:07Z",
          "capture": "superaris_readme.raw",
          "capture_sha256": "b5daa2063dbad39d2cc72f4dcd4efe6e000181835c90f854e2ed6a3059fec6fd"
        }
      ],
      "reported_facts": [
        "The public MIT repository and README describe a Python research workbench built on ARIS, with a coordinating main role, persistent engineer/writer teammates, task records, and project notes.",
        "The current README labels the project preview v0.1.0, but GitHub's releases and tags endpoints were empty at the audit observation."
      ],
      "publisher_interpretation": [
        "New Runtime presents persistent roles and recoverable project notes as a shift from one-shot research prompts to ongoing multi-step work."
      ],
      "analyst_inference": [
        {
          "text": "The design is relevant to durable agent tasks and evidence trails, but the v0.1.0 label is not a tagged release and the current README revision postdates the RSS entry.",
          "label": "inference"
        }
      ],
      "forecast_claims": [],
      "result": "partially_confirmed",
      "limitations": [
        "First-party repository text verifies the project description, not successful autonomous operation, result quality, or acceptance safeguards.",
        "Current README commit was published after the 13:00:03Z RSS timestamp, so these exact details cannot all be back-attributed to the selected publication time."
      ],
      "contradictions": [
        "The article calls this v0.1.0 preview; no Git tag or GitHub release for v0.1.0 was found. The README badge alone is not a release artifact."
      ],
      "recheck_condition": "Pin a release tag and dated source snapshot before citing current features; verify task completion, evidence checks, cancellation, and recovery behavior independently.",
      "duplicate_flags": []
    },
    {
      "selection_index": 10,
      "claim_id": "NR-20261109-10",
      "title": "EAGLE-4 0.1.2 falls back to plain SGLang decoding at high load and posts TokenSpeed tokens/s",
      "newruntime_url": "https://newruntime.com/posts/eagle-4-0-1-2-falls-back-to-4d2aee/",
      "publisher_type": "field_note",
      "rss_published_at_utc": "2026-11-08T09:00:02Z",
      "primary_event_date": {
        "value": "2026-10-23T11:21:39Z",
        "basis": "GitHub commit daeb41e documents 0.1.2 images; prior 0.1.2 behavior change commit df5a23a is 2026-10-23T01:36:47Z. Version 0.1.3 was released by commit e110d1e on 2026-10-27T03:19:47Z."
      },
      "primary_sources": [
        {
          "url": "https://github.com/eagle-hq/EAGLE-4/blob/main/docs/release-notes/0.1.2.md",
          "kind": "official_release_notes",
          "version": "0.1.2",
          "source_date": "2026-10-23",
          "source_commit": "daeb41ecd2fcdc6524db3bf559cfe386152135cc",
          "capture": "eagle_notes.raw",
          "capture_sha256": "ebcb247fe41c1887e0952e2c5596a43b132e776870137e969a8fa14c6624c5bc"
        },
        {
          "url": "https://github.com/eagle-hq/EAGLE-4/blob/main/docs/release-notes/0.1.3.md",
          "kind": "later_official_release_notes",
          "version": "0.1.3",
          "source_commit": "e110d1e916518ad9af145e84fc524891314fd78a",
          "source_date": "2026-10-27T03:19:47Z",
          "capture": "eagle_release_0.1.3.raw",
          "capture_sha256": "ddf4f2114251d2bbeaca7f9810eb8d186c7dbbdaa01f583566f4942d7887bad7"
        }
      ],
      "reported_facts": [
        "The 0.1.2 notes describe per-batch SGLang fallback to plain decoding and list measured throughput conditions.",
        "Those same notes state the TokenSpeed engine build from that release was withdrawn; the current 0.1.3 notes describe a later TokenSpeed release."
      ],
      "publisher_interpretation": [
        "New Runtime summarizes a near-parity high-load SGLang recipe and low-load speedups; the raw release notes say this requires a four-GPU setup and note the ordinary one-GPU quick-start can be slower above roughly 32–64 concurrency."
      ],
      "analyst_inference": [
        {
          "text": "The 0.1.2 mechanism is historically supported, but the item is stale as of the Nov 8 publication because 0.1.3 had shipped on Oct 27; EAGLE-4 is an inference-serving system, not an agent harness or message-delivery system.",
          "label": "inference"
        }
      ],
      "forecast_claims": [],
      "result": "stale",
      "limitations": [
        "Benchmark values are vendor release-note claims, not independently reproduced here.",
        "TokenSpeed data must not be read as an available 0.1.2 engine build; the release explicitly says that build was withdrawn."
      ],
      "contradictions": [
        "The selected article reports 0.1.2 on Nov 8 despite an official 0.1.3 release on Oct 27; the high-load mechanism may still be present but the version focus is stale."
      ],
      "recheck_condition": "Use current release 0.1.3 or newer and its own support matrix; rerun the benchmark only with matched hardware, load, engine, and no-spec baseline.",
      "duplicate_flags": []
    }
  ]
}
