Proposal: add a versioned public-safe JSON export for downstream tools
Nessuno ha ancora preso questa issue.
Valutazione
- Difficoltà
- 5/5
- Tempo stimato
- Più di una settimana
- Idoneità per principianti
- 38/100
- Tipo di issue
- Funzionalità
- Chiarezza
- Abbastanza chiara
- Stato di attività
- Attiva
- Stack tecnologico
- javascript
- Ambito
- tooling
Direzione di ricerca
Start by locating the existing full JSON export, the VERSION value, and the representations for completed lanes and probe metadata. Agree the public-safe envelope with maintainers, then add an offline test for the version, metadata, selected results, and excluded fields; document the non-authentication and non-safety-rating boundary.
Scritto dal modello di indicizzazione a partire dal testo della issue.
Descrizione
Context
ModelPrint currently exports completed lanes as an unversioned JSON array:
[
{
"provider": "openrouter",
"model": "example/model",
"results": {},
"raw": {}
}
]
The existing export is useful for local forensic analysis: raw intentionally preserves timing, headers, request IDs, and error details. This proposal would keep that export available and add a separate, versioned public-safe export for downstream tools.
This is an interoperability/format proposal, not a vulnerability report.
Consumer use case
I am exploring optional interoperability with API Relay Audit, following the documentation contribution merged in ModelPrint PR #4.
API Relay Audit would treat a ModelPrint artifact as independent, informational model-fingerprint evidence. It would not turn fingerprint matches into an audit verdict, include them in its 6D risk matrix, imply ModelPrint endorsement, or claim that shared infrastructure proves model identity.
Proposal
Add a second export path, such as Public-safe JSON, with a small versioned envelope. For example:
{
"schema_version": "modelprint-export-v1",
"tool_version": "0.7.4",
"generated_at": "2026-08-24T00:00:00Z",
"export_mode": "public-safe",
"probes": [
{"id": "tok-english", "version": "1.0.0"}
],
"lanes": [
{
"provider": "openrouter",
"model": "example/model",
"results": {
"tok-english": 42
}
}
],
"raw_included": false
}
The exact field names are open to maintainer preference; the important interoperability properties are an explicit schema version, the ModelPrint version, probe IDs plus versions, and a public-safe lane/result collection.
Suggested acceptance criteria
- Keep the current full forensic JSON export available.
- Add a distinct public-safe export path.
- Include a schema version, ModelPrint version, generation timestamp, and selected probe versions.
- Export only completed lanes and selected probes' deterministic
valueresults. - Exclude
raw, response headers, request IDs, timing samples, error bodies, custom endpoints, and pinned hosts. - Add an offline test covering the envelope, consistency with
VERSION, probe metadata, and absence of excluded fields. - Document that fingerprints are infrastructure/model-family signals, not identity authentication or a safety rating.
If this boundary fits ModelPrint's direction, I can follow up with an implementation PR after the schema shape is agreed.
- Lingua principale
- JavaScript
- Stelle
- 119
- Fork
- 14
- Merge medio
- 10h 23m
- PR unite (30g)
- 4
Guida per i contributori
Apri la guida per i contributori
Come iniziare
- Leggi tutta la issue e poi la guida ai contributi del progetto.
- Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
- Fai un fork del repository e lavora su un branch.
- Apri una pull request che faccia riferimento al numero della issue.
Issue simili
-
bug customer-eng Durable Agents Inngest status: needs triage
Difficoltà 2/5 1-3 ore Idoneità per principianti 82/100
-
optimization optimization:agents-md-curator
Difficoltà 2/5 1-3 ore Idoneità per principianti 75/100
githubnext/gh-aw-cao#13475 ·
-
[BUG]: "Clear All" in Settings doesn't clear the saved analysis, old data comes back after reload Apertabug
Difficoltà 2/5 1-3 ore Idoneità per principianti 88/100
AOSSIE-Org/OrgExplorer#253 · 1 commento ·
-
Difficoltà 2/5 1-3 ore Idoneità per principianti 88/100
oxc-project/oxc#26944 ·
-
ai-observability bug team/ai-observability
Difficoltà 2/5 1-3 ore Idoneità per principianti 78/100