Fleet Schemas ↑↓ move within a pane · ←→ move between panes · Enter opens · Esc backs out

Closed enum registry

Research Link Sources

Every allowed value of links[].source, its optional URL rule, and its meaning. This table is read from the executable research-records schema metadata.

SourceClassURL patternMetadata schemaDescription
arxiv-paper platform ^https://arxiv\.org/abs/\d{4}\.\d{4,5}$
{
  "description": "Provider-backed arXiv absence or artifact result.",
  "anyOf": [
    {
      "description": "Provider-backed absence of an arXiv preprint for this Paper.",
      "type": "object",
      "properties": {
        "unavailable": {
          "description": "What the provider could not supply and the only discriminator of this metadata arm, such as pdf; a provider-specific marker key in its place is banned.",
          "type": "string",
          "const": "preprint"
        },
        "reason": {
          "description": "The provider-typed cause of that absence drawn from this subject's closed vocabulary, such as closed; free text or a cause invented at the call site is banned.",
          "type": "string",
          "enum": [
            "arxiv-search-miss"
          ]
        },
        "recheck": {
          "description": "Where this absence sits in the recheck cycle: the arXiv search snapshot that last looked, such as snapshot-v298, or final once the paper is over 18 months old and is never re-checked; any other value is banned.",
          "type": "string",
          "pattern": "^(snapshot-v\\d+|final)$"
        }
      },
      "required": [
        "unavailable",
        "reason",
        "recheck"
      ],
      "additionalProperties": false
    },
    {
      "description": "Pointer to a query-arxiv-owned PDF artifact, validated when it was saved.",
      "type": "object",
      "properties": {
        "artifactPath": {
          "description": "Absolute path into that provider's own artifact store, such as <provider-store>/papers/<id>.pdf; a relative path or a record-local copy is banned.",
          "type": "string",
          "minLength": 1
        }
      },
      "required": [
        "artifactPath"
      ],
      "additionalProperties": false
    }
  ]
}
arXiv paper identity anchor.
doi platform ^https://doi\.org/10\.\d+/.+$
{
  "description": "Provider-backed DOI acquisition result.",
  "anyOf": [
    {
      "description": "Provider-backed absence of usable PDF bytes for a DOI Paper, whether no source was recorded, the selected host refused them, or every open-access source failed and the mirror fallback refused (mirror-refused).",
      "type": "object",
      "properties": {
        "unavailable": {
          "description": "What the provider could not supply and the only discriminator of this metadata arm, such as pdf; a provider-specific marker key in its place is banned.",
          "type": "string",
          "const": "pdf"
        },
        "reason": {
          "description": "The provider-typed cause of that absence drawn from this subject's closed vocabulary, such as closed; free text or a cause invented at the call site is banned.",
          "type": "string",
          "enum": [
            "closed",
            "no-direct-pdf",
            "doi-not-found",
            "pdf-host-http-403",
            "mirror-refused"
          ]
        }
      },
      "required": [
        "unavailable",
        "reason"
      ],
      "additionalProperties": false
    },
    {
      "description": "Pointer to the validated full text proving a DOI Paper: a query-unpaywall PDF, or a site-warc capture of the publisher's own page when provider is site-warc.",
      "type": "object",
      "properties": {
        "artifactPath": {
          "description": "Absolute path into that provider's own artifact store, such as <provider-store>/papers/<id>.pdf; a relative path or a record-local copy is banned.",
          "type": "string",
          "minLength": 1
        },
        "provider": {
          "description": "Set only when the publisher serves no PDF anywhere (Distill is the case): the pointer then names a site-warc capture of the publisher's own article page instead of a query-unpaywall PDF, accepted only after every PDF route failed. Absent means the ordinary query-unpaywall PDF proof.",
          "type": "string",
          "const": "site-warc"
        }
      },
      "required": [
        "artifactPath"
      ],
      "additionalProperties": false
    }
  ]
}
DOI-resolved publication identity anchor.
openreview-forum platform ^https://openreview\.net/forum\?id=.+$
{
  "description": "Pointer to a query-openreview-owned PDF artifact, validated when it was saved.",
  "type": "object",
  "properties": {
    "artifactPath": {
      "description": "Absolute path into that provider's own artifact store, such as <provider-store>/papers/<id>.pdf; a relative path or a record-local copy is banned.",
      "type": "string",
      "minLength": 1
    }
  },
  "required": [
    "artifactPath"
  ],
  "additionalProperties": false
}
OpenReview forum identity anchor; carries the query-openreview PDF pointer only for a Paper with no arXiv or DOI identity.
openreview-profile platform ^https://openreview\.net/profile\?id=~.+$ - OpenReview researcher profile (one per person, self-managed).
openalex-author platform ^https://openalex\.org/A\d+$
{
  "description": "Provider-backed absence needed to keep an affiliation-less Person visibly blocked.",
  "type": "object",
  "properties": {
    "unavailable": {
      "description": "What the provider could not supply and the only discriminator of this metadata arm, such as pdf; a provider-specific marker key in its place is banned.",
      "type": "string",
      "const": "institution"
    },
    "reason": {
      "description": "The provider-typed cause of that absence drawn from this subject's closed vocabulary, such as closed; free text or a cause invented at the call site is banned.",
      "type": "string",
      "enum": [
        "openalex-institution-absent"
      ]
    }
  },
  "required": [
    "unavailable",
    "reason"
  ],
  "additionalProperties": false
}
OpenAlex author record.
openalex-work platform ^https://openalex\.org/W\d+$
{
  "description": "OpenAlex-reported citation count for this paper.",
  "type": "object",
  "properties": {
    "citedByCount": {
      "description": "OpenAlex citation count observed when this surface was last read; counts differ by index, so it is never copied onto another source's row.",
      "type": "integer",
      "minimum": 0,
      "maximum": 9007199254740991
    }
  },
  "required": [
    "citedByCount"
  ],
  "additionalProperties": false
}
OpenAlex work record for this paper.
openalex-source platform ^https://openalex\.org/S\d+$ - OpenAlex source representing this venue's main publication stream, not workshops or satellite proceedings.
orcid-profile platform ^https://orcid\.org/\d{4}-\d{4}-\d{4}-\d{3}[\dX]$ - ORCID researcher profile.
semanticscholar-author platform ^https://www\.semanticscholar\.org/author/(?:[^/]+/)?\d+$ - Semantic Scholar author page (S2AND-disambiguated).
semanticscholar-paper platform ^https://www\.semanticscholar\.org/paper/(?:[^/]+/)?[0-9a-f]{40}$
{
  "description": "Semantic Scholar-reported citation count for this paper.",
  "type": "object",
  "properties": {
    "citedByCount": {
      "description": "Semantic Scholar citation count observed when this surface was last read; counts differ by index, so it is never copied onto another source's row.",
      "type": "integer",
      "minimum": 0,
      "maximum": 9007199254740991
    }
  },
  "required": [
    "citedByCount"
  ],
  "additionalProperties": false
}
Semantic Scholar paper record, the second citation-count index beside openalex-work; at most one per Paper.
x-profile platform ^https://x\.com/[^/]+/?$ - X profile.
substack-publication platform ^https://(?:www\.)?[^/]+\.substack\.com/?$ - Substack publication.
wikipedia-page platform ^https://[^/]+\.wikipedia\.org/wiki/.+$ - Wikipedia page.
linkedin-profile platform ^https://(?:www\.)?linkedin\.com/in/[^/]+/?$ - LinkedIn profile.
program-profile platform ^https://.+$ - The person's own profile page on a fellowship, mentoring, or research-program site (a mentor or fellow page such as https://www.pivotal-research.org/mentors/profile/<slug>). Not a page about the person - those are cited inline.
facebook-profile platform ^https://www\.facebook\.com/(?:profile\.php\?id=[0-9]+|(?!(?:profile\.php|people|watch|groups|pages|events|marketplace|gaming|reels|stories|login|help|settings|share|sharer|photo|photos|videos|search|home|friends|notifications)/?$)[A-Za-z0-9.]+/?)$ - Facebook identity profile for a Person, using https://www.facebook.com/<handle> or https://www.facebook.com/profile.php?id=<digits>; normalize a /people/<name>/<digits>/ address to profile.php?id=<digits> before saving. Verify the profile against the person's own site or career facts and write VERIFIED: with checkedOn; it does not supply enhanced-level corroboration.
instagram-profile platform ^https://www\.instagram\.com/(?!(?:p|reel|reels|stories|explore|accounts|direct|about|developer|legal|privacy|web)/$)[A-Za-z0-9._]+/$ - Instagram identity profile for a Person at https://www.instagram.com/<handle>/, with the trailing slash. Verify the profile against the person's own site or career facts and write VERIFIED: with checkedOn; it does not supply enhanced-level corroboration.
github-profile platform ^https://github\.com/[^/]+/?$ - GitHub profile.
huggingface-profile platform ^https://huggingface\.co/[^/]+/?$ - Hugging Face profile.
huggingface-space platform ^https://huggingface\.co/spaces/[^/]+/.+$ - Hugging Face Space containing a reproduction.
huggingface-dataset platform ^https://huggingface\.co/datasets/[^/]+/.+$ - Hugging Face dataset such as judge verdicts.
dblp-author platform ^https://dblp\.org/pid/.+$ - DBLP author profile.
google-scholar-profile platform ^https://scholar\.google\.[a-z.]+/citations\?(?:[^#]*&)?user=[A-Za-z0-9_-]+(?:[&#].*)?$ - Google Scholar author profile, any country domain (scholar.google.com, scholar.google.co.uk). A Scholar profile is self-claimed and can list a namesake's papers, so confirm the listed works are this human's before stamping it VERIFIED.
pypi-profile platform ^https://pypi\.org/user/[^/]+/?$ - PyPI user profile.
$conf-icml26 skill-event ^https://icml\.cc/virtual/2026/.+$ - ICML 2026 program evidence owned by $conf-icml26.
$mdr-skill skill-event ^https://wsh\.md/mdr/[a-z0-9-]+/SKILL\.md/$ - The mdr skill dedicated to this record, linked by its published SKILL.md page.
apple-podcasts-episode platform ^https://podcasts\.apple\.com/.+$
{
  "type": "object",
  "properties": {
    "finished": {
      "description": "Apple Podcasts has counted at least one full play of this episode, read from the user's library at the row's checkedOn. Written only as true: a play count above zero never goes back down, so absent means not finished as far as anyone has looked. This flag records completion, not its date; live play counts and playheads stay in Apple's library.",
      "type": "boolean",
      "const": true
    },
    "lastPlayedAt": {
      "description": "Apple Podcasts' last-played time for this episode, as a UTC ISO-8601 instant from the user's library. Written by the listening pass alongside finished when Apple supplies it, including for an episode already marked finished. A missing provider date leaves the stored value unchanged. This is the last play, not the time the episode was first completed; the Media page shows it as Last played.",
      "type": "string",
      "pattern": "^\\d{4}-\\d{2}-\\d{2}T\\d{2}:\\d{2}:\\d{2}(?:\\.\\d+)?Z$"
    }
  },
  "additionalProperties": false
}
Apple Podcasts episode: on a Media record, an identity of the conversation (the first such row is the original; later rows are reruns or clips of it); on a Paper, the narration of that Paper read aloud, which is never its own Media record.
apple-podcasts-show platform ^https://podcasts\.apple\.com/[a-z]{2}/podcast/(?:[^/?#]+/)?id\d+/?$ - The Apple Podcasts page of one podcast show, `https://podcasts.apple.com/<country>/podcast/[<name>/]id<number>`; the number is Apple's public show id.
youtube-channel platform ^https://www\.youtube\.com/channel/UC[A-Za-z0-9_-]{22}$ - The YouTube page of one channel by its permanent id, `https://www.youtube.com/channel/UC...`; never a handle URL, which its owner can change.
canonical-url platform ^https?://.+$
{
  "type": "object",
  "properties": {
    "artifactPath": {
      "description": "Absolute path into that provider's own artifact store, such as <provider-store>/papers/<id>.pdf; a relative path or a record-local copy is banned.",
      "type": "string",
      "minLength": 1
    },
    "provider": {
      "description": "Owner of the saved copy. query-lesswrong is permitted only for a LessWrong, Alignment Forum or EA Forum post and holds the full post followed by every API comment. site-warc, or an omitted provider on a historical row, names the retained page capture. A new save never overwrites historical evidence.",
      "type": "string",
      "enum": [
        "site-warc",
        "query-lesswrong"
      ]
    }
  },
  "required": [
    "artifactPath"
  ],
  "additionalProperties": false
}
The work's full text, which IS this record's canonical identity - the PDF when the work exists as one (a report, system card, or dissertation with no arXiv or DOI identity), otherwise the page that is the work itself (a blog or forum post); a landing page ABOUT the work (a repository record, a tech-report index, a cover sheet) is never canonical and goes in a related-url row; an http:// address only when its https:// address does not serve, with the link description carrying HTTPS-FAILED: <the observed failure>; exactly one per record, deduplicated by URL. Pages ABOUT a record are related-url rows, which are never identities. An optional metadata pointer names its saved reader Markdown: query-lesswrong for a LessWrong, Alignment Forum or EA Forum post saved through its API, site-warc for other pages or a retained historical capture.
related-url fallback ^https://.+$
{
  "type": "object",
  "properties": {
    "artifactPath": {
      "description": "Absolute path into that provider's own artifact store, such as <provider-store>/papers/<id>.pdf; a relative path or a record-local copy is banned.",
      "type": "string",
      "minLength": 1
    },
    "provider": {
      "description": "The saved-copy owner for this LessWrong, Alignment Forum or EA Forum post. The pointer includes the complete post and comment thread and never changes the Paper's canonical full text.",
      "type": "string",
      "const": "query-lesswrong"
    }
  },
  "required": [
    "artifactPath",
    "provider"
  ],
  "additionalProperties": false
}
A page about or around this record - project page, coverage; the specific relationship belongs in this row's description; a Person's own site is a homepage row instead. Never an identity, never a w.sh page (a link to another of our records belongs in a reference field). A forum crosspost may carry a query-lesswrong saved-copy pointer so chat can include its complete post and comment thread as supporting context.
transcript-url platform ^https://.+$
{
  "type": "object",
  "properties": {
    "artifactPath": {
      "description": "Absolute path into that provider's own artifact store, such as <provider-store>/papers/<id>.pdf; a relative path or a record-local copy is banned.",
      "type": "string",
      "minLength": 1
    },
    "provider": {
      "type": "string",
      "const": "site-warc"
    }
  },
  "required": [
    "artifactPath",
    "provider"
  ],
  "additionalProperties": false
}
This Media conversation's transcript page published by the show itself, at most one per record. Adding it is the choice to use it: the w.sh transcript reader prefers it over Apple and YouTube transcripts once its metadata points at a saved site-warc copy, cut to the transcript by the site's configured start and end markers; a related-url never changes the transcript.
homepage platform ^https?://.+$ - A Person's own site - their personal or research homepage, their own university profile page, or their own personal blog; at most two per Person. A lab's page goes only on the lab's principal investigator, since each address belongs to one record. A page that merely mentions them is cited inline in enhanced.why, never stored as a link.
lesswrong-profile platform ^https://www\.(?:lesswrong\.com|alignmentforum\.org)/users/[^/?#]+/?$ - A LessWrong user page, which is the same account as the person's Alignment Forum page. Many alignment researchers publish there first.
$conf-talks skill-event -
{
  "description": "conf-talks session identity for one Media document.",
  "type": "object",
  "properties": {
    "sessionId": {
      "description": "Stable conf-talks session identity defined by the provider skill.",
      "type": "string",
      "minLength": 1
    }
  },
  "required": [
    "sessionId"
  ],
  "additionalProperties": false
}
conf-talks session identity defined by the provider skill.
youtube-video platform ^https://www\.youtube\.com/watch\?v=[A-Za-z0-9_-]{11}$ - YouTube video identity anchor for a Media document. Stored in exactly one form, https://www.youtube.com/watch?v=<11-character id> with no other query parameters, so the URL is the dedupe key; the minter rewrites youtu.be links and strips extra parameters before the row is written. Proven hydrated by the query-youtube transcript artifact; the video bytes are never held.
vimeo-video platform ^https://vimeo\.com/[1-9][0-9]{0,11}$ - Vimeo video identity anchor for a Media document. Stored in exactly one form, https://vimeo.com/<numeric id> with no query, fragment, or trailing slash, so the URL is the dedupe key; the minter rewrites www.vimeo.com and player.vimeo.com links before the row is written. Proven hydrated by the query-youtube caption transcript artifact; the video bytes are never held. Only a public video is imported: an unlisted video's private link is never stored.