> ## Documentation Index > Fetch the complete documentation index at: https://docs.arrays.org/llms.txt > Use this file to discover all available pages before exploring further. # List channel commentary across people > [OpenAPI JSON Spec](/docs/output/v1_persons_channel-commentary_get.json) Same rows as /v1/persons/{person_id}/channel-commentary without fixing the person. Browse by said_on day with start_time and end_time, which returns the rows that are served now. Or poll changes with updated_after and updated_before (unix seconds, [after, before), ordered by updated_at then id, paged by pagination.cursor; see docs/api/update-polling.md and docs/api/person-channel-commentary-updates.md). An update window returns every row whose latest change falls in it, and served says which kind it is. A served row has served=true. A duplicate, which reports a remark another page holds in full, has served=false and rejected_as="duplicate", with cluster_id the original's url. A tombstone has served=false, any other rejected_as, and only the identity fields filled. A consumer deletes a row it holds when it comes back with served=false and rejected_as other than "duplicate". Poll by updated_at, never by said_on: rows arrive with old said_on days. person_id is an optional filter. No row for a person and day does not mean they said nothing: a remark is stored only when its page can be read and the sentence checked against it. ## OpenAPI ````yaml /api-reference/openapi.json get /v1/persons/channel-commentary openapi: 3.0.0 info: contact: {} description: >- Welcome to Arrays! Arrays is a unified financial data layer for both financial institutions and retail users, delivering comprehensive market data across crypto and equity markets. Our datasets cover fundamentals, ETFs, options, crypto, and more. Visit [arrays.org](https://arrays.org) to request an API key and start your free trial. You can explore our full API offerings in the documentation below. For any questions regarding APIs or pricing, please contact support@arrays.org. title: Arrays API version: '1.0' servers: - url: https://data-tools.prd.arrays.org/api security: [] paths: /v1/persons/channel-commentary: get: tags: - Persons summary: List channel commentary across people description: >- [OpenAPI JSON Spec](/docs/output/v1_persons_channel-commentary_get.json) Same rows as /v1/persons/{person_id}/channel-commentary without fixing the person. Browse by said_on day with start_time and end_time, which returns the rows that are served now. Or poll changes with updated_after and updated_before (unix seconds, [after, before), ordered by updated_at then id, paged by pagination.cursor; see docs/api/update-polling.md and docs/api/person-channel-commentary-updates.md). An update window returns every row whose latest change falls in it, and served says which kind it is. A served row has served=true. A duplicate, which reports a remark another page holds in full, has served=false and rejected_as="duplicate", with cluster_id the original's url. A tombstone has served=false, any other rejected_as, and only the identity fields filled. A consumer deletes a row it holds when it comes back with served=false and rejected_as other than "duplicate". Poll by updated_at, never by said_on: rows arrive with old said_on days. person_id is an optional filter. No row for a person and day does not mean they said nothing: a remark is stored only when its page can be read and the sentence checked against it. parameters: - description: Person registry id (uuid) to narrow to one person in: query name: person_id schema: type: string - description: >- Browse range start, unix seconds; required unless updated_after is given in: query name: start_time schema: type: integer - description: >- Browse range end, unix seconds; required unless updated_after is given in: query name: end_time schema: type: integer - description: Update window start, unix seconds, inclusive in: query name: updated_after schema: type: integer - description: Update window end, unix seconds, exclusive in: query name: updated_before schema: type: integer - description: Opaque continuation of an update window in: query name: cursor schema: type: string - description: >- default 100, or 50 with include_body; 1 to 500, at most 50 with include_body in: query name: limit schema: type: integer - description: Rows to skip; browse only in: query name: offset schema: type: integer - description: Return each row's full body in: query name: include_body schema: type: boolean responses: '200': description: OK content: application/json: schema: allOf: - $ref: '#/components/schemas/model.APIResponse' - properties: data: items: $ref: '#/components/schemas/model.PersonChannelCommentary' type: array type: object '400': description: Bad Request content: application/json: schema: $ref: '#/components/schemas/model.APIResponse' '503': description: Service Unavailable content: application/json: schema: $ref: '#/components/schemas/model.APIResponse' components: schemas: model.APIResponse: properties: data: {} error: $ref: '#/components/schemas/model.APIError' metadata: {} pagination: {} request_id: type: string success: type: boolean type: object model.PersonChannelCommentary: properties: author: description: >- Author is what the page said about who wrote it, empty when it said nothing. type: string body: description: >- Body is the page's text as read, kept as read. It is returned only when the request sets include_body=true, and limit is then at most 50 because rows that carry a page of text are large. Never returned on a tombstone. It can hold words that are not the person's, so Quote stays the sentence attributed to them. type: string body_scope: description: >- BodyScope says how complete the page text is: "full", "preview" when a paywall cut it, "truncated" when the platform or the page gave only part, "unknown" when the page does not say, "unavailable" when a re-read found the page gone. It is set whether or not Body is returned. Empty on a tombstone. type: string channel: description: >- Channel is the channel the page came from, such as "gnews", "cnbc_news" or "x_own". Empty on a tombstone. type: string cluster_id: description: >- ClusterID is the row's own url on a served row, and the url of the original on a duplicate row, which is how a report is found under its original. Empty on a tombstone. type: string context: description: >- Context holds the posts this one answers, quotes or forwards, each with its own author. It is other people's words and never part of Quote or Body. An empty array, never null, when there are none, and on a tombstone. items: $ref: '#/components/schemas/model.PersonCommentaryContext' type: array fetched_at: description: >- FetchedAt is when the page was retrieved, unix seconds. 0 on a tombstone. type: integer id: description: >- ID is the row's fixed identity, derived from person_id, said_on and url, so it stays the same when the row is rewritten. It is the same derivation as the id of /api/v1/persons/commentary, and an update window is ordered by updated_at, then id. type: string lang: type: string original_at: description: >- OriginalAt is set only on a duplicate row: the "[mm:ss]" position of the quote in the original's captions or transcript. Absent when the position is unknown. type: string own_channel_verified: description: >- OwnChannelVerified is set only for "own": "author" when the page named this person; "channel" when the page is theirs to publish but is not shown to be written by them; "unverified" when neither could answer. type: string page_title: description: >- PageTitle and Venue describe where it was said. Venue is the program, publication or platform. type: string person_id: description: >- PersonID is the person registry's id, the same one /api/v1/persons answers with. Resolve a name to one with /api/v1/persons?name=... first. type: string published_at: description: >- PublishedAt is unix seconds, from the page itself unless PublishedAtSource says otherwise. 0 on a tombstone. type: integer published_at_source: description: >- PublishedAtSource is "page" when the page stated the time and "fetch" when it stated none and the retrieval time stands in. Empty on a tombstone. type: string quote: description: >- Quote is one sentence, character for character as the page prints it. Empty on a tombstone. type: string rejected_as: description: >- RejectedAs is absent on a served row. It is "duplicate" on a row that reports another page's remark. On a tombstone it is the code of why the row is not served: the code of a refusal, or "firsthand_report" for a report still waiting for its original to be found. type: string reply_status: description: >- ReplyStatus is "not_reply" when the row does not answer another post, "context_complete" when it does and every post it answers is in Context, "context_incomplete" when one of those could not be read (its status says why), and "unknown" for a page of a kind whose replies are not read. Empty on a tombstone. type: string said_at: description: >- SaidAt is when the remark was said, unix seconds. Absent when SaidAtSource is "unknown", and on a tombstone. type: integer said_at_source: description: >- SaidAtSource is where SaidAt came from: "post" for the post's own timestamp, "page_stated" for a date the page prints, which names a day and not a time, and "unknown" when neither could be established. Empty on a tombstone. type: string said_on: description: >- SaidOn is the UTC day this row is filed under, the day the remark was said; see said_at for when it was said. type: string served: description: >- Served is true when the row is output, a row to keep. A consumer that holds a row whose served is false and whose rejected_as is not "duplicate" should delete it. type: boolean source_kind: description: >- SourceKind is "own", "transcript", "third_party_transcript" or "interview". "transcript" is the venue's own transcript; "third_party_transcript" is one another site published with the person's name on it. Empty on a tombstone. type: string speaker_label: description: >- SpeakerLabel is the transcript's own label on the passage the quote is in, such as "CRAMER". Empty when the page names no speaker. type: string statement_level: description: >- StatementLevel is "original" when the page is where they said or published it, "firsthand_report" when a reporter heard them directly and the report is the fullest text of it, "unknown" when it has not been judged. Empty on a tombstone. type: string text_origin: description: >- TextOrigin is what Body was made from: "page" for the page's own text, "captions_manual" for captions a person wrote, "captions_asr" for speech recognition. Machine transcripts can mishear words. Empty on a tombstone. type: string updated_at: description: >- UpdatedAt is when this row last changed, as RFC3339 UTC with fractional seconds. A change of state is a change: updated_after and updated_before on /api/v1/persons/channel-commentary filter on it. type: string url: description: >- URL is the address the text was read from, after the fetcher resolved it. type: string venue: type: string type: object model.APIError: properties: code: example: RESOURCE_NOT_FOUND type: string details: items: $ref: '#/components/schemas/model.APIErrorDetail' type: array docs_url: type: string examples: items: type: string type: array hint: type: string message: example: The requested resource was not found. type: string suggestions: items: type: string type: array type: object model.PersonCommentaryContext: properties: author_handle: type: string author_id: description: >- AuthorID, AuthorHandle and AuthorName are what the platform gave for the post's author, each left out when it gave none. type: string author_name: type: string depth: description: >- Depth is 1 for the post the row answers directly, then 2 and 3 for the posts above it. type: integer published_at: description: PublishedAt is unix seconds, 0 when the platform gave no time. type: integer relation: description: >- Relation is "reply_to", "quoted", "forwarded", "reshared" or "thread_root". type: string status: description: >- Status is "ok", "deleted" or "unavailable". Anything but "ok" means the post could not be read in full, and the row's ReplyStatus says so. type: string text: description: Text is the post's text as the platform gave it. type: string url: description: URL is the address of this post. type: string type: object model.APIErrorDetail: properties: field: type: string got: type: string reason: type: string type: object ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.