Create a linkedin posts agent

Free text matched against the POST TEXT. Each value is trimmed and lowercased, then all values are OR-joined into one provider query (["hiring", "humanoids"] is sent as "hiring OR humanoids"), so a post matching ANY one of them qualifies; at least one keyword is required and there is no cap. A single value may itself carry boolean syntax: uppercase AND, OR and NOT, parentheses for grouping and double quotes for exact phrases, for example [""product manager" AND hiring"].

Size of the posting company, as a headcount bucket. Use one of exactly: 1-10, 11-50, 51-200, 201-500, 501-1,000, 1,001-5,000, 5,001-10,000, 10,001+ (the commas and the exact spacing are part of the value: "500-1000" is not a bucket). The upstream stores the string as sent and never checks it against a list, so any other value is accepted, the create still answers 201, and the agent then silently matches nothing. Multiple buckets are OR-joined (a company in any listed band matches) and there is no cap. At least one of companyHeadcount or companyCountries must be set before this agent can start.

Headquarters country of the posting company. Use values from the provider's canonical country list: full English country names such as "United States", "United Kingdom", "Germany". ISO codes like "US" are not accepted, no lookup endpoint serves this vocabulary yet, and an unrecognised name is accepted, the create still answers 201, and the agent then silently matches nothing. Multiple countries are OR-joined (a company headquartered in any of them matches) and there is no cap. At least one of companyHeadcount or companyCountries must be set before this agent can start.

Industry of the person or company posting. Use exact values from GET /business/agents/autocomplete?field=industry (for example "Software Development", "Financial Services"). Any other wording is first resolved to the closest supported LinkedIn industry ("tech" becomes "Technology, Information and Internet"), and a value with no match is dropped rather than rejected, so the filter silently widens instead of failing. At most 1 value is allowed here; a second entry is rejected with 422 VALIDATION_FAILED.

Companies the post author currently works at, as company LinkedIn URLs, for example ["https://www.linkedin.com/company/google/"]. This filter resolves URLs only: a bare company name or domain is accepted, the create still answers 201, and the agent then silently matches nothing. Multiple URLs are OR-joined (a post by an employee of any listed company matches) and there is no cap. Preferred over the deprecated currentCompany.

Current job title of the post author. Use exact values from GET /business/agents/autocomplete?field=title (for example "Vice President of Sales", "CEO"). The upstream stores the string as sent and never checks it against a list, so any other value is accepted, the create still answers 201, and the agent then silently matches nothing. Multiple titles are OR-joined (a post by anyone holding any of them matches) and there is no cap. Preferred over the deprecated currentTitle.

Author location, matched geographically; at most one value, e.g. ["United States"]. Use exact values from GET /business/agents/autocomplete?field=region.

Whether to match posts from people, companies or both (the default). At most one entry each of "person" and "company".

A single plain-language filter on the POST CONTENT itself, e.g. ["announcing a new product launch"]. It describes the post, never the person or company posting.

Categories the post must fall into, for example ["Business", "Personal", "Events", "Social Commentary"]. Free text: the provider classifies every post with its own labels and files anything unmatched under "Others", so there is no fixed list and no endpoint to look one up, and a label it does not use is accepted, the create still answers 201, and the agent then silently matches nothing. Multiple categories are OR-joined (a post in any of them matches); the provider caps this filter at 10 categories, and a longer array is accepted here and rejected upstream.

Deprecated, use authorCompany. Read only when authorCompany is omitted, and then sent as the author-company filter, which resolves company LinkedIn URLs only, for example "https://www.linkedin.com/company/google/": the bare name or domain this field is named after is accepted, the create still answers 201, and the agent then silently matches nothing. Multiple URLs are OR-joined and there is no cap.

Past employer of the post author, as a company NAME or DOMAIN, for example ["Google"] or ["google.com"]. Free text resolved by the provider, so a name it cannot resolve is accepted, the create still answers 201, and the agent then silently matches nothing; a domain is the unambiguous form. Multiple values are OR-joined (an author who worked at any of them matches) and there is no cap.

Deprecated, use authorTitle. Read only when authorTitle is omitted, and then sent as the author-title filter. Use exact values from GET /business/agents/autocomplete?field=title (for example "Vice President of Sales"). The upstream stores the string as sent and never checks it against a list, so any other value is accepted, the create still answers 201, and the agent then silently matches nothing. Multiple titles are OR-joined and there is no cap.

A job title anywhere in the post author's PAST work history. Use exact values from GET /business/agents/autocomplete?field=title (for example "Vice President of Sales", "CEO"). The upstream stores the string as sent and never checks it against a list, so any other value is accepted, the create still answers 201, and the agent then silently matches nothing. Multiple titles are OR-joined (an author who held any of them matches) and there is no cap.

Size of the post author's CURRENT employer, as a headcount bucket. Use one of exactly: 1-10, 11-50, 51-200, 201-500, 501-1,000, 1,001-5,000, 5,001-10,000, 10,001+ (the commas and the exact spacing are part of the value: "500-1000" is not a bucket). The upstream stores the string as sent and never checks it against a list, so any other value is accepted, the create still answers 201, and the agent then silently matches nothing. Multiple buckets are OR-joined (an author whose employer is in any listed band matches) and there is no cap.

Headquarters country of the post author's CURRENT employer. Use values from the provider's canonical country list: full English country names such as "United States", "United Kingdom", "Germany". ISO codes like "US" are not accepted, no lookup endpoint serves this vocabulary yet, and an unrecognised name is accepted, the create still answers 201, and the agent then silently matches nothing. Multiple countries are OR-joined (an author whose employer sits in any of them matches) and there is no cap.