Workspace · local-first

Your own classes, appraisals and evidence packs

Bring a crossing class, appraise proposals against it, record what happened afterwards, replay every determination and observation from the file you keep, and export a reproducible evidence pack. It all happens in this browser tab.

Access

Method, synthetic examples, proof surfaces, tests, non-claims	public	Available on this site
Local-first workspace: your own class, appraisal, replay, evidence pack	public	This page; runs in your browser
Pilot: a private class configured with you, review, and a signed-off evidence pack	proposed	An engagement using this workspace; not yet offered
Hosted workspaces, accounts, collaboration	proposed	Not built
Subscriptions	proposed	Not built
Human sign-off on an evidence pack	proposed	Not built
Recognition of packs by an insurer or regulator	proposed	None exists

Payment, where it is offered, would change access only. It never changes a verdict, a warrant, scientific standing, certification or cover.

Local-first: classes, proposals, determinations, observations and packs stay in this page and in files you download. This page makes no network requests while you use it. Close the tab without saving and the workspace is gone.

1. Workspace

NameOpen a workspace file

0 class(es) admitted · 0 admission record(s) · 0 determination(s) · 0 observation(s) · sha256:d7ce8de0277b1e45…

An opened file is checked in full first. If anything is malformed, it is rejected with every problem listed and the open workspace stays as it was.

2. Bring your own crossing class

Paste a class written against the class schema. The science class is preloaded as a starting point. A warranted example lets the mutation battery run; without one the battery is reported unavailable, not passed. Admission is syntactic: it does not validate the class for its domain or show that its consequence paths are complete.

Class JSON

{
  "class_version": "nt-crossing-class:v0",
  "id": "science.claim-promotion",
  "version": "0.1.0",
  "title": "Promotion of an experimental result to a claim",
  "domain": "empirical research",
  "evidence_status": "proposed",
  "summary": "A result moves from 'reported' to 'claimed with standing' so others may reuse it. The path is instrument → mapping → analysis → interpretation → claim → reuse. A missing or disputed measurement is value uncertainty; an analysis path that was never executed from authoritative input challenges the structure; competing mappings of the same evidence end UNRESOLVED. Not an insurance class.",
  "state_fields": {
    "before": [
      "experiment_id",
      "result_id",
      "analysis_path"
    ],
    "proposed": [
      "claim_id",
      "claim",
      "mappings"
    ]
  },
  "action_types": [
    "promote_claim"
  ],
  "policy": {
    "id": "nt-policy:claim-promotion",
    "version": "0.1.0",
    "authored_by": "Neverthought (illustrative class policy; not a journal's or funder's standard)",
    "parameters": {
      "max_delegation_depth": 1
    },
    "note": "Promotion needs the raw results, an independent reproduction, and a reviewer other than the builder."
  },
  "invariants": [
    {
      "id": "mapping_agreement",
      "label": "Every declared mapping of the evidence reports the same value",
      "measure": "mappings_agree",
      "params": {
        "field": "mappings"
      },
      "tolerance": 0,
      "register": "value"
    },
    {
      "id": "analysis_path",
      "label": "The analysis path the claim relies on was executed from authoritative input",
      "measure": "path_executable",
      "params": {
        "field": "analysis_path"
      },
      "tolerance": 0,
      "register": "structure"
    }
  ],
  "evidence_requirements": [
    {
      "subject": "raw_results",
      "min_tier": "E2",
      "required_to_resolve": "Preserve the raw results, including the pre-state needed to reconstruct each comparison from bytes on disk."
    },
    {
      "subject": "independent_reproduction",
      "min_tier": "E4",
      "required_to_resolve": "Obtain a reproduction by a party outside the builder's authority."
    }
  ],
  "authority_requirements": {
    "independent_trust_root": true,
    "segregation_of_duties": true
  },
  "archetypes": [
    "RA-SELF-CERTIFICATION",
    "RA-HIDDEN-COUPLING",
    "RA-UNWARRANTED-EXECUTION"
  ],
  "descriptors": {
    "stake": "reputational",
    "action": "compensable",
    "authority_kind": "explicit",
    "freshness": "perpetual"
  },
  "profiles": {
    "recorder": "nt-receipt-chain:v0 (this site's chain; not CER-17)",
    "underwriting": null
  },
  "consequence": {
    "bound": "unbounded",
    "ceiling": null,
    "rationale": "A claim with standing is reused by others; errors propagate through every reuse, so no ceiling is declared."
  },
  "consequence_paths": [
    {
      "id": "claim_reused",
      "description": "Others build on the claim."
    },
    {
      "id": "claim_retracted",
      "description": "The claim is withdrawn or corrected."
    },
    {
      "id": "replication_attempted",
      "description": "An independent party attempts to reproduce the result."
    }
  ],
  "structural_tests": [],
  "failure_mode": {
    "mode": "fail-closed",
    "rationale": "A result that is held is not presented as a claim."
  },
  "non_claims": [
    "Promotion records that a result met this class's checks; it does not certify a discovery or establish causality.",
    "The worked examples map a review of software-governance experiments, not a natural-science study.",
    "Methodological inference in the source review is recorded as not adjudicated, and stays so here."
  ]
}

Warranted example proposal (optional, for the battery)

{
  "proposal_version": "nt-proposal:v0",
  "proposal_id": "SCI-001",
  "class_id": "science.claim-promotion",
  "class_version": "0.1.0",
  "domain": "empirical research",
  "s_before": {
    "experiment_id": "EPISTEMIC_WIND_TUNNEL_MCP_EXPERIMENT_SLICE_001",
    "result_id": "SURF_14",
    "analysis_path": {
      "executed": true,
      "from_authoritative_input": true
    },
    "source": {
      "review": "Documents/experiment contract discovery/CODEX_EMPIRICAL_REVIEW_RESUMED_v0.1.0",
      "file": "SURF_14_ABLATION_REEXECUTION.md",
      "sha256": "3b45c209af2d01b3f16e1b12ac819b27a6fadf0a197788bea98663c2b1ffa7fa"
    }
  },
  "action": {
    "type": "promote_claim",
    "purpose": "Promote the result to a claim others may reuse",
    "effects": [
      {
        "target": "claim-register",
        "description": "status reported to claimed"
      }
    ]
  },
  "s_proposed": {
    "claim_id": "CLAIM-SCI-001",
    "claim": "Removing the bundle-type guard loses exactly four of five malformed-shape detections",
    "mappings": [
      {
        "mapping_id": "builder report",
        "value": "4 of 5"
      },
      {
        "mapping_id": "independent re-execution",
        "value": "4 of 5"
      }
    ]
  },
  "t_valid": "2026-09-20T12:00:00Z",
  "authority": {
    "principal": "claim-review-agent",
    "trust_root": {
      "id": "independent-review-authority",
      "declared_by": "review commissioner, outside the builder's authority",
      "independent_of_proposer": true
    },
    "delegation_chain": [
      {
        "from": "review-commissioner",
        "to": "claim-review-agent",
        "scope": "claim promotion"
      }
    ],
    "preparer": "experiment-builder",
    "approver": "independent-reviewer"
  },
  "evidence": [
    {
      "id": "E1",
      "subject": "raw_results",
      "source": "builder run log",
      "sha256": "d25a4e08962ef590f9ef9068199e0aa604e4b002296a9fcd2da1d751e9310d27",
      "captured_at": "2026-09-15T09:00:00Z",
      "valid_until": null,
      "binds_to": "SCI-001",
      "tier": "E2"
    },
    {
      "id": "E2",
      "subject": "independent_reproduction",
      "source": "independent reviewer",
      "sha256": "5683decd7202bbc822d4dcd85d1239ecd8c24159257682b528b5f42912487fb0",
      "captured_at": "2026-09-15T09:00:00Z",
      "valid_until": null,
      "binds_to": "SCI-001",
      "tier": "E4"
    }
  ],
  "dependencies": {
    "declared": [
      "claim-register"
    ],
    "data_source": "wind-tunnel-slice-001"
  },
  "execution_state": "proposed"
}

3. Appraise proposals

Class

4. Record consequence observations

Record what happened after a crossing: the original input, its source, when it was observed, and the path you think it followed. The workspace tries that path against the crossing’s frozen class and keeps the attempt, matched or not. It never adds a path to the class. Value status (known, pending, disputed) and structure status (untested, challenged) are kept apart. An unmatched observation challenges the structure only when you record why its source is independent; that basis is declared, not verified. Without one, it is kept as unresolved evidence. Corrections and supersessions are new records that point back; nothing earlier is rewritten. Schema: observation-input.v0.

Crossing

Bounded omission detection: only observations supplied here are checked against the frozen class. An effect absent from every supplied observation cannot be detected. Source independence is declared, not verified. No class path is created from an observation.

5. Replay from the file

6. Evidence pack

Verify a pack

A matching manifest shows these files are the ones listed. It does not show the set is complete, that the inputs are true, or who produced the pack.
Hashes and self-replay establish internal consistency only. They do not establish source authenticity, that an observed consequence was caused by the crossing, or that the class's consequence model is complete.
Determinations and observation standings are self-replayed from the workspace's own bytes; this is not independent replay.
Source independence on an observation is declared by the observer, not verified. An effect absent from every supplied observation cannot be detected.
Classes carry syntactic admission only; they are not validated for their domain.
No verdict, warrant, scientific standing, certification or insurance cover is conferred by this pack or by payment for it.