v1

Search Summarizer

Produce a single, concise summary over the full corpus of articles matching your filters, using your prompt to guide which insights to highlight.

post/v1/summarize

Query parameters

qstring

Primary search query for filtering articles based on their title, description, and content. Supports Boolean operators (AND, OR, NOT), exact phrases with quotes, and wildcards (* and ?) for flexible searching.

titlestring

Search specifically within article headlines/titles. Supports Boolean operators, exact phrases with quotes, and wildcards for matching title variations.

descstring

Search within article description fields. Supports Boolean expressions, exact phrase matching with quotes, and wildcards for flexible pattern matching.

contentstring

Search within the full article body content. Supports Boolean logic, exact phrase matching with quotes, and wildcards for comprehensive content searching.

urlstring

Search within article URLs to find content from specific website sections or domains. Supports wildcards (* and ?) for partial URL matching.

articleIdstring[]

Retrieve specific news articles by their unique article identifiers. Multiple IDs can be provided to return a collection of specific articles.

clusterIdstring[]

Filter results to only show content within a specific related content cluster. Returns articles grouped together as part of Perigon Stories based on topic relevance.

sortBy'relevance' | 'date' | 'reverseDate' | 'reverseAddDate' | 'addDate' | 'pubDate' | 'refreshDate'

Determines the article sorting order. Options include relevance (default), date/pubDate (newest publication date first), reverseDate (oldest publication date first), addDate (newest ingestion date first), reverseAddDate (oldest ingestion date first), and refreshDate (most recently updated in system first, often identical to addDate).

pageinteger

The specific page of results to retrieve in the paginated response. Starts at 0.

sizeinteger

The number of articles to return per page in the paginated response.

fromstring date-time

Filter for articles published after this date. Accepts ISO 8601 format (e.g., 2023-03-01T00:00:00) or yyyy-mm-dd format.

tostring date-time

Filter for articles published before this date. Accepts ISO 8601 format (e.g., 2022-02-01T23:59:59) or yyyy-mm-dd format.

addDateFromstring date-time

Filter for articles added to Perigon's system after this date. Accepts ISO 8601 format (e.g., 2022-02-01T00:00:00) or yyyy-mm-dd format.

addDateTostring date-time

Filter for articles added to Perigon's system before this date. Accepts ISO 8601 format (e.g., 2022-02-01T23:59:59) or yyyy-mm-dd format.

refreshDateFromstring date-time

Filter for articles refreshed/updated in Perigon's system after this date. In most cases yields similar results to addDateFrom but can differ for updated content. Accepts ISO 8601 format (e.g., 2022-02-01T00:00:00) or yyyy-mm-dd format.

refreshDateTostring date-time

Filter for articles refreshed/updated in Perigon's system before this date. In most cases yields similar results to addDateTo but can differ for updated content. Accepts ISO 8601 format (e.g., 2022-02-01T23:59:59) or yyyy-mm-dd format.

mediumstring[]

Filter articles by their primary medium type. Accepts Article for written content or Video for video-based stories. Multiple values create an OR filter.

sourcestring[]

Filter articles by specific publisher domains or subdomains. Supports wildcards (* and ?) for pattern matching (e.g., *.cnn.com). Multiple values create an OR filter.

sourceGroupstring[]

Filter articles using Perigon's curated publisher bundles (e.g., top100, top25crypto). Multiple values create an OR filter to include articles from any of the specified bundles.

excludeSourceGroupstring[]

Exclude articles from specified Perigon source groups. Multiple values create an AND-exclude filter, removing content from publishers in any of the specified bundles (e.g., top10, top100).

excludeSourcestring[]

Exclude articles from specific publisher domains or subdomains. Supports wildcards (* and ?) for pattern matching (e.g., *.cnn.com). Multiple values create an AND-exclude filter.

paywallboolean

Filter to show only results where the source has a paywall (true) or does not have a paywall (false).

bylinestring[]

Filter articles by author bylines. Works as an exact match for each author name provided. Multiple values create an OR filter to find articles by any of the specified authors.

authorstring[]

Filter articles by specific author names. Works as an exact match for each name. Multiple values create an OR filter to find articles by any of the specified authors.

excludeAuthorstring[]

Exclude articles written by specific authors. Any article with an author name matching an entry in this list will be omitted from results. Multiple values create an AND-exclude filter.

journalistIdstring[]

Filter by unique journalist identifiers which can be found through the Journalist API or in the matchedAuthors field. Multiple values create an OR filter.

excludeJournalistIdstring[]

Exclude articles written by specific journalists identified by their unique IDs. Multiple values create an AND-exclude filter.

languagestring[]

Filter articles by their language using ISO-639 two-letter codes (e.g., en, es, fr). Multiple values create an OR filter.

excludeLanguagestring[]

Exclude articles in specific languages using ISO-639 two-letter codes. Multiple values create an AND-exclude filter.

searchTranslationboolean

Expand search to include translated content fields for non-English articles. When true, searches translated title, description, and content fields.

labelstring[]

Filter articles by editorial labels such as Opinion, Paid-news, Non-news, Fact Check, or Press Release. Multiple values create an OR filter.

excludeLabelstring[]

Exclude articles with specific editorial labels. Multiple values create an AND-exclude filter, removing all content with any of these labels.

categorystring[]

Filter by broad content categories such as Politics, Tech, Sports, Business, or Finance. Use 'none' to find uncategorized articles. Multiple values create an OR filter.

excludeCategorystring[]

Exclude articles with specific categories. Multiple values create an AND-exclude filter, removing all content with any of these categories.

topicstring[]

Filter by specific topics such as Markets, Crime, Cryptocurrency, or College Sports. Topics are more granular than categories, and articles can have multiple topics. Use the /topics endpoint for a complete list of available topics. Multiple values create an OR filter.

excludeTopicstring[]

Exclude articles with specific topics. Multiple values create an AND-exclude filter, removing all content with any of these topics.

linkTostring

Returns only articles that contain links to the specified URL pattern. Matches against the 'links' field in article responses.

showReprintsboolean

Controls whether to include reprinted content in results. When true (default), shows syndicated articles from wire services like AP or Reuters that appear on multiple sites.

reprintGroupIdstring

Returns all articles in a specific reprint group, including the original article and all its known reprints. Use when you want to see all versions of the same content.

citystring[]

Filters articles where a specified city plays a central role in the content, beyond mere mentions, to ensure the results are deeply relevant to the urban area in question. If multiple parameters are passed, they will be applied as OR operations.

excludeCitystring[]

A list of cities to exclude from the results. Articles that are associated with any of the specified cities will be filtered out.

areastring[]

Filters articles where a specified area, such as a neighborhood, borough, or district, plays a central role in the content, beyond mere mentions, to ensure the results are deeply relevant to the area in question. If multiple parameters are passed, they will be applied as OR operations.

statestring[]

Filters articles where a specified state plays a central role in the content, beyond mere mentions, to ensure the results are deeply relevant to the state in question. If multiple parameters are passed, they will be applied as OR operations.

excludeStatestring[]

A list of states to exclude. Articles that include, or are associated with, any of the states provided here will be filtered out. This is especially useful if you want to ignore news tied to certain geographical areas (e.g., US states).

countystring[]

A list of counties to include (or specify) in the search results. This field filters the returned articles based on the county associated with the event or news. Only articles tagged with one of these counties will be included.

excludeCountystring[]

Excludes articles from specific counties or administrative divisions in the search results. Accepts either a single county name or a list of county names. County names should match the format used in article metadata (e.g., 'Los Angeles County', 'Cook County'). This parameter allows for more granular geographic filter

locationsCountrystring[]

Filters articles where a specified country plays a central role in the content, beyond mere mentions, to ensure the results are deeply relevant to the country in question. If multiple parameters are passed, they will be applied as OR operations.

countrystring[]

Country code to filter by country. If multiple parameters are passed, they will be applied as OR operations.

excludeLocationsCountrystring[]

Excludes articles where a specified country plays a central role in the content, ensuring results are not deeply relevant to the country in question. If multiple parameters are passed, they will be applied as AND operations, excluding articles relevant to any of the specified countries.

locationstring[]

Return all articles that have the specified location. Location attributes are delimited by ':' between key and value, and '::' between attributes. Example: 'city:New York::state:NY'.

latnumber double

Latitude of the center point to search places

lonnumber double

Longitude of the center point to search places

maxDistancenumber double

Maximum distance (in km) from starting point to search articles by tagged places

sourceCitystring[]

Find articles published by sources that are located within a given city.

sourceCountystring[]

Find articles published by sources that are located within a given county.

sourceCountrystring[]

Find articles published by sources that are located within a given country. Must be 2 character country code (i.e. us, gb, etc).

sourceStatestring[]

Find articles published by sources that are located within a given state.

sourceLatnumber double

Latitude of the center point to search articles created by local publications.

sourceLonnumber double

Latitude of the center point to search articles created by local publications.

sourceMaxDistancenumber double

Maximum distance from starting point to search articles created by local publications.

personWikidataIdstring[]

Filter articles by Wikidata IDs of mentioned people. Refer to the /people endpoint for a complete list of tracked individuals.

excludePersonWikidataIdstring[]

Exclude articles mentioning people with specific Wikidata IDs. Creates an AND-exclude filter to remove content about these individuals. Uses precise identifiers to avoid name ambiguity.

personNamestring[]

Filter articles by exact person name matches. Does not support Boolean or complex logic. For available person entities, consult the /people endpoint.

excludePersonNamestring[]

Exclude articles mentioning specific people by name. Creates an AND-exclude filter to remove content about these individuals.

companyIdstring[]

Filter articles by company identifiers. For a complete list of tracked companies, refer to the /companies endpoint.

excludeCompanyIdstring[]

Exclude articles mentioning companies with specific identifiers. Creates an AND-exclude filter to remove content about these corporate entities.

companyNamestring

Filter articles by company name mentions. Performs an exact match on company names.

companyDomainstring[]

Filter articles by company domains (e.g., apple.com). For available company entities, consult the /companies endpoint.

excludeCompanyDomainstring[]

Exclude articles related to companies with specific domains. Creates an AND-exclude filter to remove content about these companies.

companySymbolstring[]

Filter articles by company stock symbols. For available company entities and their symbols, consult the /companies endpoint.

excludeCompanySymbolstring[]

A list of stock symbols (ticker symbols) that identify companies to be excluded. Articles related to companies using any of these symbols will be omitted, which is useful for targeting or avoiding specific public companies.

showNumResultsboolean

Whether to show the total number of all matched articles. Default value is false which makes queries a bit more efficient but also counts up to 10000 articles.

positiveSentimentFromnumber float

Filter articles with a positive sentiment score greater than or equal to the specified value. Scores range from 0 to 1, with higher values indicating stronger positive tone.

positiveSentimentTonumber float

Filter articles with a positive sentiment score less than or equal to the specified value. Scores range from 0 to 1, with higher values indicating stronger positive tone.

neutralSentimentFromnumber float

Filter articles with a neutral sentiment score greater than or equal to the specified value. Scores range from 0 to 1, with higher values indicating stronger neutral tone.

neutralSentimentTonumber float

Filter articles with a neutral sentiment score less than or equal to the specified value. Scores range from 0 to 1, with higher values indicating stronger neutral tone.

negativeSentimentFromnumber float

Filter articles with a negative sentiment score greater than or equal to the specified value. Scores range from 0 to 1, with higher values indicating stronger negative tone.

negativeSentimentTonumber float

Filter articles with a negative sentiment score less than or equal to the specified value. Scores range from 0 to 1, with higher values indicating stronger negative tone.

taxonomystring[]

Filters by Google Content Categories. This field will accept 1 or more categories, must pass the full name of the category. Example: taxonomy=/Finance/Banking/Other, /Finance/Investing/Funds. Full list

prefixTaxonomystring

Filters by Google Content Categories. This field will filter by the category prefix only. Example: prefixTaxonomy=/Finance

showHighlightingboolean

When set to true, enables text highlighting in search results.

highlightFragmentSizeinteger

Specifies the size in characters of each highlighted text fragment. Defaults to 100 if not specified.

highlightNumFragmentsinteger

Controls the maximum number of highlighted fragments to return per field.

highlightPreTagstring

Defines the HTML tag that appears before highlighted text. Defaults to '<em>' if not specified.

highlightPostTagstring

Defines the HTML tag that appears after highlighted text. Defaults to '</em>' if not specified.

highlightQstring

Specifies a separate query for highlighting, allowing highlights based on terms different from the main search query. Example: main query 'q=climate change' with 'highlightQ=renewable OR solar' will highlight terms 'renewable' and 'solar' in results about climate change.

Request body

promptstring nullable

Instructions guiding how the summary should be written. Maximum length: 2 048 characters.

maxArticleCountinteger nullable

The maximum number of articles to factor into the summary.

returnedArticleCountinteger nullable

The maximum number of articles that should be returned in the response. This can be used to return fewer than maxArticleCount results.

summarizeFields'TITLE' | 'CONTENT' | 'SUMMARY' nullable

Which article fields to include when generating the summary. Up to three values from TITLE, CONTENT, SUMMARY.

method'ARTICLES' | 'CLUSTERS' nullable

Method for selecting articles: ARTICLES (include all matches) or CLUSTERS (one per cluster).

model'gpt-4o' | 'gpt-4o-mini' | 'gpt-4.1' | 'gpt-4.1-mini' | 'gpt-4.1-nano' | 'llama-3.3-70b-versatile' | 'deepseek-r1-distill-llama-70b' nullable

The underlying LLM model to use for generation.

temperaturenumber double nullable

Sampling temperature for the LLM (0.0 = deterministic to 2.0 = very creative).

topPnumber double nullable

Nucleus sampling (top-p) for the LLM (0.0 to 1.0).

maxTokensinteger nullable

Maximum number of tokens to generate in the summary.

Response

Successful summary response

statusinteger required
numResultsinteger required
summarystring required

Changes