Tools · Find people by LinkedIn engagement
search_intent_people
Find who wrote, commented on or reacted to LinkedIn posts on your topic, filtered to your persona.
When to use
Use it for people who showed intent on LinkedIn: who posted, commented on or reacted to posts about a topic, to given posts, or to the recent posts of given people or companies, optionally only those matching a persona ("product people at airlines who engaged with posts about ChatGPT apps").
Don’t use it to find people by role alone (use search_people).
Request
curl https://lead-search-production.up.railway.app/v1/search_intent_people \
-H "Authorization: Bearer $MUNCH_KEY" \
-d '{"topic": "AI SDR", "keywords": ["AI SDR", "outbound agent"],
"titles": ["VP Sales", "Head of RevOps"]}'
Over MCP, your agent fills these parameters from your request.
Parameters
All optional. Give at least one search criterion.
qstringWho the people must be, in one plain sentence: role + kind of company + place, e.g. "marketing or product people at airlines".
actionsstring[]Intent that counts. Default all: "post" (wrote an on-topic post the search found: the strongest intent, listed first), "comment", "react".
post comment reactaffiliatedstringPeople whose engagement may be loyalty, not intent: they work where the post's author or most of its audience works, or the post is about their employer (staff liking their own launch).
exclude label includeauthorsstring[]LinkedIn pages (https://www.linkedin.com/in/<name> or /company/<name>) whose recent posts to read instead of searching, e.g. a competitor, an industry voice.
clarifystring"auto" (default): when you has not said what counts and the posts found are of types that would change who is found, the answer returns `post_types` with `needs_answer` and no people, so you can choose.
auto nevercompaniesstring[]Only people at these companies (domains, names or LinkedIn company pages), e.g. ["airfrance.com", "Lufthansa"].
company_descriptionstringThe kind of company, e.g. "airlines and travel companies": each employer is checked against it.
confirm_employersbooleantrue (default): each returned person’s LinkedIn profile is read to confirm their current employer (~$0.0005 each). false: only people whose headline names no employer are read.
cursorstring`next_cursor` from the previous answer, with exactly the same criteria (`relevant` / `not_relevant` may be added or changed): reads the next engagers without searching or paying again for what was already read.
exclude_companiesstring[]People to leave out by employer (names, domains or LinkedIn company pages): customers, competitors, companies already contacted, e.g. ["airfrance.com", "Lufthansa"].
industriesstring[]LinkedIn industries of the persona's companies, e.g. ["Airlines and Aviation"]: posts by authors in them are searched first (their readers are the persona's peers), then all posts.
keywordsstring[]Searches for posts on the topic (default: the topic).
limitintegerPeople wanted in total, over all pages; a call often gives 4-8 and says `found_so_far`.
not_relevantstring[]What does not count, in your words (e.g. ["agencies selling TikTok Shop services"]) or type ids from `post_types`.
per_companyintegerMost people per employer in the answer (0: no cap), so one company's staff cannot fill it.
persona_fillbooleantrue (default): when on-topic posts run out before `limit`, more people are found by role and company and their own recent activity is checked for the topic (~$0.006 a person). false: posts only.
postsstring[]LinkedIn post URLs whose commenters and reactors to read instead of searching, e.g. a competitor's launch post. Best first: they are read in this order.
relevantstring[]What counts as the intent, in your own words, e.g. ["brands sharing their own TikTok Shop launch or results", "founders asking peers about it"]; or, after an answer with `post_types`, the ids of the types that count (e.g. ["g1", "g4"]).
response_formatstring"concise" (default): the fields needed to pick and act on rows. "detailed": every field (LinkedIn URLs, descriptions, industries); about 3-5x the tokens.
concise detailedsince_daysintegerHow far back intent counts, in days (posts read, and the persona's own activity).
titlesstring[]Roles wanted and their variants, e.g. ["CMO", "Head of Marketing", "Head of Product"]: engagers whose headline names none of these jobs are dropped first (free).
topicstringWhat the posts are about, in a few words, e.g. "launching an app in ChatGPT".
Response
budget_left_today_usdnumberWhat this key may still spend today (daily cap, resets 00:00 UTC); absent when uncapped
cost_usdnumberfound_so_farintegerPeople returned over all pages so far, this page included; while it is below `limit` and next_cursor is set, call again with the cursor
needs_answerbooleanTrue when the user must first say which `post_types` count: no people yet. Ask them, then call again with next_cursor and relevant / not_relevant
next_cursorstringPass back as `cursor` with the same criteria to read more without paying again for what was read
notesstring[]Plain-language notes for the user: what was found, what could not be confirmed, and what to do next
peopleIntentRow[]post_typesPostType[]The types of post found on the topic and whether each counts as intent ('ask': only the user can say), with real examples; on the call that looked at them
postsPostRow[]Detailed answer: the posts read, with the people each gave
people[] fields
companystringcompany_domainstringemployer_confirmedbooleanDetailed answer: true when `company` is their current job on their LinkedIn profile; false when read from their headline
intentany[]What they did on which post
last_intent_datestringTheir most recent action on the topic (YYYY-MM-DD), or the date of the post they reacted to
linkedin_urlstringlocationstringnamestringperson_idstringPass to find_contacts to get this person's email or phone
summarystringTheir headline, as found
tierstringDetailed answer: 'intent' (wrote, commented on or reacted to posts on the topic), 'persona_with_intent' (found by role and company once the posts ran dry; their own recent activity touches the topic), 'persona_no_intent' (found by role and company, no public activity on the topic) or 'persona_unchecked' (found by role and company, activity not looked at in time)
titlestringpost_types[] fields
aboutstringThe topic itself, or the wider field around it
countedstringWhether engaging with such posts counts as intent: 'ask' when only the user can say
doesstringWhat such posts do, e.g. 'sharing their own launch, adoption or results'
examplesany[]Real posts of this type
idstringHow the user's answer names this type: put it in `relevant` or `not_relevant`
sharenumberShare of the posts found on the topic that are of this type (0-1)
written_byobjectWho writes such posts, by share, e.g. {"vendors and agencies": 0.6}
posts[] fields
authorstringdatestringkindstring'first_hand' (the author or their company launched, built or uses it), 'discussion' (news, opinion) or 'promo' (selling or teaching it)
peopleintegerPeople in this answer who engaged with it
textstringThe post's first words
urlstring