Creator
Drafting a post
Claude, help me write a post about LinkedIn's algorithm change. Pull my top 3 posts on algorithm topics so you see my voice, and score the draft before we publish.
LinkedIn intelligence for AI agents
LinkedIn publishes no official MCP server that we could find. Flux is an independent, hosted one: connect Claude, Claude Code, Cursor, or any MCP client, then ask about any public profile or company page. You never share a LinkedIn login or cookie, and there is nothing to install.
Free preview
Flux’s preview tools, preview_profile and preview_posting_insights, analyze up to 50 recent posts from a public profile or company page. Add https://www.fluxgraph.app/mcp to your client, then ask a question like the one below. Usage limits apply, and some results may require a free Flux account.
Use Flux to preview linkedin.com/in/janedoe. Which days, hours, formats, and lengths beat this account’s own baseline, and how many posts back each finding?
Compare
The main difference is how a server reaches LinkedIn. That decides whose account is exposed, what it can see, and what it costs.
| Question | Your logged-in session | LinkedIn API with OAuth | Flux (hosted) |
|---|---|---|---|
| How it reaches LinkedIn | Automates a browser or reuses your cookie | LinkedIn’s API, under your authorization | A third-party provider of public data |
| Whose account is exposed | Yours | Yours, within the permissions you grant | Not yours, because you connect no LinkedIn account |
| Other people’s posts | Whatever your session can see | Very limited | Any public profile or company page |
| Stored history and trends | Usually none | Usually none | Yes, on paid plans |
| Can post or message | Often | Sometimes (posting) | No, read-only on LinkedIn |
| Setup | Local install and upkeep | Your own LinkedIn developer app, usually | Paste one URL |
| Cost | Usually free software | Usually free software | Free 50-post preview, then paid plans |
This table describes common designs, and individual servers vary. As of November 2026, Flux’s price pays for the data pipeline and stored history, not for the protocol.
Why MCP matters
Flux ships one. Most days you won’t open it. Ask Claudewhat’s working in your last quarter, where engagement broke, who to reach out to next — it queries your post library, embeddings, scoring model, and constellation graph and writes the answer. Your data lives one sentence away from whatever you’re already working on.
Vignettes
Writing, researching, and synthesizing — each backed by your own Flux data instead of whatever Claude happens to guess.
Creator
Claude, help me write a post about LinkedIn's algorithm change. Pull my top 3 posts on algorithm topics so you see my voice, and score the draft before we publish.
BDR
Map the constellation for Acme Corp. Who are the senior execs, which of my connections is closest to each, and what does their recent posting look like?
Consultant
Find my 10 most-engaged posts about SaaS pricing. Summarize the core thesis across them.
Setup · 2 minutes
One endpoint, https://www.fluxgraph.app/mcp, two ways in. Claude connects by signing in with your Flux account. Developer tools connect with an MCP key. Open your client below, or read the step-by-step connection guide.
Sign in with Flux · no key
Claude — web, desktop app, mobile
Sign inAdd Flux as a custom connector. Connectors sync across claude.ai, the desktop app, and mobile.
https://www.fluxgraph.app/mcp.Developer tools · MCP key
Claude Code
MCP keyOne terminal command adds Flux for every project.
flux_ and is shown once; the same panel shows this snippet with your key filled in.YOUR_API_KEY:claude mcp add --transport http --scope user flux \ https://www.fluxgraph.app/mcp \ --header "Authorization: Bearer YOUR_API_KEY"
Why a key: Claude Code picks a random local port for browser sign-in, and Flux’s sign-in server needs an exact one.
Cursor
MCP keyAdd a server entry to ~/.cursor/mcp.json, or .cursor/mcp.json for one project.
flux_ and is shown once; the same panel shows this snippet with your key filled in.~/.cursor/mcp.json (all projects) or .cursor/mcp.json (one project):{
"mcpServers": {
"flux": {
"url": "https://www.fluxgraph.app/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Zed
MCP keyAdd a context server to settings.json, or use Settings → AI → MCP Servers.
flux_ and is shown once; the same panel shows this snippet with your key filled in.settings.json:{
"context_servers": {
"flux": {
"url": "https://www.fluxgraph.app/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Other MCP clients
MCP keyAnything that supports HTTP transport and custom headers.
flux_ and is shown once; the same panel shows this snippet with your key filled in.{
"mcpServers": {
"flux": {
"type": "http",
"url": "https://www.fluxgraph.app/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Calling the server from your own code? See Calling tools directly (HTTP) below.
Tool reference
Free tools show free; paid tools call out their credit cost. Open a group to see each tool’s input fields, return keys, and any tool-specific errors. The JSON-RPC envelope and common error codes live in the first section below, and the full parameter reference is in the MCP tool reference docs. To see the tools at work, start with MCP stories: real jobs, step by step, with prompts you can copy.
Calling tools directly (HTTP)
JSON-RPC 2.0 envelope, endpoint, and the errors every tool can return.
Most users will connect Claude or an MCP client like Claude Code and never touch HTTP. If you need to call the server directly, every tool uses the same JSON-RPC envelope — only name and arguments change per call.
Endpoint
POST https://www.fluxgraph.app/mcp Authorization: Bearer flux_xxx Content-Type: application/json
Envelope
{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/call",
"params": {
"name": "<tool_name>",
"arguments": { /* tool inputs */ }
}
}Example: list tracked profiles
curl -X POST https://www.fluxgraph.app/mcp \
-H "Authorization: Bearer $FLUX_API_KEY" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call",
"params":{"name":"list_profiles","arguments":{}}}'Common errors (apply to every tool)
401 invalid_api_key — key missing, malformed, or revoked. Get a fresh key at /settings.402 insufficient_credits — balance too low for a paid tool. Call get_cost_estimate first to preview spend.403 forbidden — resource exists but isn't owned by the authenticated user (most often profile_id or collection_id).422 validation_failed — arguments didn't match the tool's input schema.429 rate_limited — too many requests in the burst window. Back off and retry.Per-tool blocks below only call out errors that are specific to that tool (e.g. insufficient_data, 404 not_found). The five above are universal — assume they can return on any call.
Preview
Analyze any public LinkedIn profile or company page before you have a plan. preview_profile needs no account; the other three need a free Flux sign-in.
preview_profilefreePull the latest ~50 posts of a public profile or company page (once per profile) and return a teaser: post count, average engagement, the format that beats the baseline, the top post, and the recent trend, plus a preview_id. No account needed.
Use Flux to preview linkedin.com/in/janedoe.
Input
| field | type | req | notes |
|---|---|---|---|
| linkedin_url | string | yes | Public LinkedIn profile or company URL |
Returns
preview_id — stringteaser metrics — post count, average engagement, best format, top post, trendcontinue_url — string — opens the full breakdown after a free sign-inpreview_posting_insightsfreeThe full breakdown for a previewed profile: best day, hour, format and length, the top and bottom 5 posts with measured reasons, and engagement trend by format. Needs a free Flux account.
Show me the full breakdown for that preview.
preview_search_postsfreeKeyword search over the latest 50 posts of a previewed profile, ordered by engagement. Needs a free Flux account.
Which of those posts mention pricing?
claim_previewfreeMake a previewed profile your own free Flux dashboard. Use it only for your own profile: a claim cannot be undone, and an account that already has a dashboard is refused.
That profile is mine. Set it up as my dashboard.
Analytics
Query your own post history — KPIs, trends, patterns, and semantic/keyword search.
get_postsfreeReturn posts with full metrics and text content for a tracked profile. Supports filtering by date range, post type, minimum engagement, and text search. Pass scope to restrict to a collection. Paginated.
Show me every post I published last quarter with more than 50 likes.
Input
| field | type | req | notes |
|---|---|---|---|
| profile_id | string | yes | Tracked profile ID from list_profiles |
| scope | string | no | Optional post_collections.id to filter |
| date_from | string | no | Start date (ISO 8601) |
| date_to | string | no | End date (ISO 8601) |
| post_type | enum | no | text, image, video, document, article, poll |
| min_engagement | number | no | Minimum total engagement |
| search | string | no | Text search across post content |
| sort_by | enum | no | default: published_at |
| sort_dir | enum | no | default: desc |
| page | number | no | default: 1 |
| page_size | number | no | default: 25 (1–50) |
Returns
posts — Array — post objects with id, metrics, text contentmeta — Object — total_count, page, page_size, total_pages, filters_appliedErrors
404 — Collection not found (when scope provided)get_profile_summaryfreeKPI summary for a tracked profile: total posts, avg engagement (likes/comments/shares/conversation rate), posts per week, engagement trend, top format, and posting goal status.
How am I doing this month vs last month?
Input
| field | type | req | notes |
|---|---|---|---|
| profile_id | string | yes | Tracked profile ID from list_profiles |
Returns
total_posts — numberavg_engagement — numberavg_likes — numberavg_comments — numberavg_shares — numberconversation_rate — number — (comments / likes) * 100posts_per_week — numberdate_range — Object — earliest, latestengagement_trend — Object — direction, change_pct, 30d averagestop_format — Object | nullposting_goal — Object — goal_per_week, posts_this_week, on_tracklinks — Object (optional)get_posting_insightsfreeOptimal posting patterns for a profile: best day of week, best hour, format performance breakdown, character-length sweet spot. All metrics relative to the user's own baseline. Pass scope to recompute against a collection.
When should I post this week to maximize comments?
Input
| field | type | req | notes |
|---|---|---|---|
| profile_id | string | yes | Tracked profile ID from list_profiles |
| scope | string | no | Optional post_collections.id to scope analysis |
| date_from | string | no | Start date (ISO 8601) |
| date_to | string | no | End date (ISO 8601) |
Returns
sample_size — number — total posts analyzedoverall_avg_engagement — numberbest_day — Object | nullbest_hour — Object | nullbest_format — Object | nullsweet_spot_length — Object | nullby_day_of_week — Array — sorted by avg_engagementby_hour — Array — sorted by avg_engagementby_format — Arrayby_length — Array — 200-char binsmeta — ObjectErrors
403 — Profile or scope not found/not ownedget_engagement_trendsfreeEngagement time series with gap-filled buckets, rolling averages, and trend direction. Each bucket includes avg engagement, likes, comments, shares, and conversation rate.
Plot my engagement over the last 12 weeks and tell me where it broke.
Input
| field | type | req | notes |
|---|---|---|---|
| profile_id | string | yes | Tracked profile ID from list_profiles |
| scope | string | no | Optional post_collections.id |
| granularity | enum | no | default: weekly (daily, weekly, monthly) |
| date_from | string | no | Start date (ISO 8601) |
| date_to | string | no | End date (ISO 8601) |
| rolling_window | number | no | 2–30 buckets; default by granularity |
Returns
series — Array — per-bucket metrics with rolling averagessummary — Object — direction, change_pct, period_comparisonmeta — Object — granularity, rolling_window, sample_sizeErrors
403 — Profile or scope not found/not ownedget_topic_insightsfreeTop terms and bigrams across a profile's posts, ranked by post count and engagement vs the user's baseline. Each term carries a sample post URL so you can jump to a representative example. Pass scope to recompute against a collection.
What topics get me the most engagement this quarter?
Input
| field | type | req | notes |
|---|---|---|---|
| profile_id | string | yes | Tracked profile ID from list_profiles |
| n | number | no | default: 15 (1–100 max terms per list) |
| bigrams | boolean | no | default: true |
| scope | string | no | Optional post_collections.id |
| date_from | string | no | Start date (ISO 8601) |
| date_to | string | no | End date (ISO 8601) |
Returns
terms — Array — term, post_count, avg_engagement, vs_baseline, sample_post_urlbigrams — Array | nulltotal_posts — numberoverall_avg_engagement — numberwindow — Object — from, tometa — ObjectErrors
403 — Profile or scope not found/not ownedsearch_postsfreeSearch posts by keyword and/or semantic similarity. Full-text search with stemming for keyword matches, pgvector embeddings for semantic matches. Results ranked by relevance.
Find everything I've written about pricing strategy.
Input
| field | type | req | notes |
|---|---|---|---|
| profile_id | string | yes | Tracked profile ID from list_profiles |
| query | string | yes | Search query |
| scope | string | no | Optional post_collections.id |
| mode | enum | no | default: hybrid (hybrid, keyword, semantic) |
| limit | number | no | default: 20 (1–50) |
Returns
posts — Array — matched posts (with similarity scores in semantic modes)meta — Object — query, mode, result_count, scopeErrors
403 — Profile or scope not found/not owned400 — Query required, or embedding generation failedfind_similar_postsfreeFind posts semantically similar to input text using OpenAI embeddings and pgvector cosine similarity. Scope to a profile or a collection. Returns posts sorted by similarity with engagement metrics.
Have I already written something like this draft?
Input
| field | type | req | notes |
|---|---|---|---|
| text | string | yes | Text to find similar posts for (min 1 char) |
| profile_id | string | no | Optional — scope to this profile's posts |
| collection_id | string | no | Optional — scope to this collection |
| limit | number | no | default: 5 (1–50) |
Returns
posts — Array — sorted by embedding similaritycount — numbercost — number — 0 (free)Errors
403 — Profile or collection not found/not owned400 — Failed to generate embeddingPrediction & drafting
Score drafts before you publish and generate fresh angles grounded in your top performers.
score_draftfreePredict how a draft post will perform against the user's historical data. OLS regression with detrending, ridge regularization, and recency weighting — same model as the /score page. Returns predicted engagement percentile, projected range, factor breakdown, similar past posts, and improvement suggestions.
Score this draft and tell me what to cut.
Input
| field | type | req | notes |
|---|---|---|---|
| text | string | yes | Draft post text (min 1 char) |
| format | enum | no | text | image | video | document | article | poll (default: text) |
| day_of_week | number | yes | 0–6, 0=Sunday (UTC) |
| hour | number | yes | 0–23 (UTC) |
| profile_id | string | no | Optional — scope model to this profile |
| collection_id | string | no | Optional — scope model to this collection |
Returns
percentile — number — predicted engagement percentile (0–100)stars — string — qualitative bandrange — Object — low/high projected engagementfactors — Array — per-feature contributionsimilar_posts — Array — nearest historical posts with their percentilesuggestions — Array — actionable improvementsmodel — Object — sample size, R², calibrationErrors
insufficient_data — not enough historical posts in scope to train; no charge.suggest_angles5 creditsGiven a topic, suggest distinct content angles informed by what performs well in the user's scoped data. Pulls top historical posts as context and asks OpenAI to generate 5 concrete angles.
Give me five angles on developer tooling that fit my voice.
Streams
Always-on research streams — external profiles, companies, topics, hiring signals, ad watches — that pull on a schedule into your library.
list_profilesfreeList all tracked LinkedIn profiles for the authenticated user. Returns profile IDs needed by other tools.
List my tracked LinkedIn profiles.
Returns
profiles — Array — id, display_name, linkedin_url, plan_type, status, posting_goal_per_week, autopilot_active, linksmeta — Object — countErrors
401 — Unauthorized (missing userId)list_streamsfreeList all research streams (external profiles, companies, topics, person interests, hiring signals, ad watches, own_profile). Pass type to filter.
Show me every stream I've got running.
Input
| field | type | req | notes |
|---|---|---|---|
| type | enum | no | own_profile, external_profile, company, topic, person_interests, hiring_signals, ad_watch |
| include_inactive | boolean | no | default: false |
Returns
streams — Array — id, type, name, source_url, params, schedule, active, credits_per_run, last/next run, linksmeta — Object — countlinks — Object (optional) — view_allcreate_streamfreeCreate a new research stream — the create call itself is free. Each scheduled run costs credits per page walked — normally one page; see get_cost_estimate for per-type cost. own_profile streams come from subscription checkout, not this tool.
Start a topic stream that watches LinkedIn for posts about RAG.
get_stream_feedfreeReturn the most recent posts a stream has collected into the library, newest first by published_at.
What did my 'fintech execs' stream pick up this week?
Input
| field | type | req | notes |
|---|---|---|---|
| stream_id | string | yes | Stream ID from list_streams |
| limit | number | no | default: 20 (1–50) |
Returns
stream — Object — id, type, nameposts — Array — recent posts (newest first)meta — Object — count, limitlinks — Object (optional) — viewErrors
404 — Stream not found or not owned by youpause_streamfreePause a stream so it stops running on its cron schedule. Existing posts remain in the library.
Pause the competitor-ads stream.
Input
| field | type | req | notes |
|---|---|---|---|
| stream_id | string | yes | Stream ID from list_streams |
Returns
stream — Object — id, type, name, active (will be false)Errors
404 — Stream not found or not owned by youresume_streamfreeResume a paused stream. Queues it to run on the next cron tick.
Resume my hiring-signals stream.
Input
| field | type | req | notes |
|---|---|---|---|
| stream_id | string | yes | Stream ID from list_streams |
Returns
stream — Object — id, type, name, active (will be true), next_run_atErrors
404 — Stream not found or not owned by youdelete_streamfreeDelete a stream and its runs. Posts it collected stay in your library so historical data isn't lost. own_profile streams can't be deleted here — cancel the subscription instead.
Drop the stream I set up for Acme Corp.
Input
| field | type | req | notes |
|---|---|---|---|
| stream_id | string | yes | Stream ID from list_streams |
Returns
deleted — boolean — true on successstream_id — stringErrors
404 — Stream not found or not owned by you403 — own_profile streams cannot be deletedrun_stream_now1–2 creditsManually trigger a stream run instead of waiting for its cron tick. Charges per page actually walked (1 credit/page for most types, 2 for person_interests), not per page requested. Useful when you want fresh results before a scheduled run.
Run my hiring-signals stream now.
Collections & Segments
Collections group posts; segments group people. Both can be smart (predicate-driven) or manual, and both can be resolved into concrete IDs for downstream tools.
list_collectionsfreeList all post collections for the authenticated user, with member counts.
Input
| field | type | req | notes |
|---|---|---|---|
| type | enum | no | manual, smart, auto_stream |
Returns
collections — Array — id, name, type, config, stream_id, created_at, member_count, linksmeta — Object — countlinks — Object (optional) — view_allcreate_collectionfreeCreate a new post collection. Manual or smart (predicate-based).
Make a collection of every post I wrote about hiring.
Input
| field | type | req | notes |
|---|---|---|---|
| name | string | yes | 1–200 chars |
| description | string | no | max 2000 chars |
| type | enum | yes | manual, smart |
| config | object | no | Predicate (required when type=smart) |
Returns
collection — Object — id, name, type, config, stream_id, created_at, member_count: 0, linksErrors
400 — config required when type=smartadd_to_collectionfreeAdd one or more posts (by post_id) to a collection. Existing members are silently ignored.
Input
| field | type | req | notes |
|---|---|---|---|
| collection_id | string | yes | Target collection (UUID) |
| post_ids | array | yes | Post UUIDs to add (1–1000 per call) |
Returns
collection_id — stringpinned — number — posts newly pinned to the collectionskipped — number — already members (deduped)member_count — number — total after operationresolved_count — number — total posts the collection now resolves to (rule matches ∪ pinned)Errors
404 — Collection not found or not owned by youremove_from_collectionfreeRemove one or more posts (by post_id) from a collection. Missing members are silently ignored.
Input
| field | type | req | notes |
|---|---|---|---|
| collection_id | string | yes | Target collection (UUID) |
| post_ids | array | yes | Post UUIDs to remove (1–1000 per call) |
Returns
collection_id — stringremoved — number — members deletednot_found — number — not membersmember_count — number — total after operationErrors
404 — Collection not found or not owned by youresolve_collectionfreeResolve a post collection into a concrete list of post_ids. Works for both manual and smart collections.
Input
| field | type | req | notes |
|---|---|---|---|
| collection_id | string | yes | Collection (UUID) |
| limit | number | no | default: 1000 (1–5000) |
Returns
collection_id — stringtype — string — manual, smart, or auto_streampost_ids — Arraycount — numberconfig — Object (smart collections only)warnings — Array — predicate-evaluation warningslinks — Object (optional) — viewErrors
404 — Collection not found or not owned by youlist_segmentsfreeList all audience segments for the authenticated user, with member counts. Segments are named groups of people (e.g. 'ICP — VPs of Marketing', 'Champions').
Input
| field | type | req | notes |
|---|---|---|---|
| type | enum | no | manual, smart, auto_stream |
Returns
segments — Array — id, name, description, type, config, source_stream_id, created_at, member_count, linksmeta — Object — countlinks — Object (optional) — view_allcreate_segmentfreeCreate a new audience segment. Manual or smart (template-based).
Create a smart segment of the VPs among my engagers who work at a bank.
Input
| field | type | req | notes |
|---|---|---|---|
| name | string | no | 1–200 chars; optional if template supplied |
| description | string | no | max 2000 chars |
| type | enum | no | manual, smart, auto_stream; defaults to smart with template |
| template | enum | no | Pre-built template id (pre-fills config) |
| config | object | no | title_contains and/or company_contains: each a string or an array of 1–20 strings (any term matches; case-insensitive substring). Both keys must match. Required when type=smart without a template |
| source_stream_id | string | no | Optional source stream UUID |
Returns
segment — Object — id, name, description, type, config, source_stream_id, created_at, member_count: 0, linksErrors
400 — name/type required (or pass template); config required when type=smart404 — source_stream_id or template id not foundadd_to_segmentfreeAdd one or more people (by person_id) to an audience segment.
Input
| field | type | req | notes |
|---|---|---|---|
| segment_id | string | yes | Target segment (UUID) |
| person_ids | array | yes | Person UUIDs to add (1–1000 per call) |
Returns
segment_id — stringadded — number — new members insertedskipped — number — already members (deduped)member_count — numberErrors
404 — Segment not found or not owned by youremove_from_segmentfreeRemove one or more people (by person_id) from an audience segment.
Input
| field | type | req | notes |
|---|---|---|---|
| segment_id | string | yes | Target segment (UUID) |
| person_ids | array | yes | Person UUIDs to remove (1–1000 per call) |
Returns
segment_id — stringremoved — number — members deletednot_found — number — not membersmember_count — numberErrors
404 — Segment not found or not owned by youresolve_segmentfreeResolve an audience segment into a concrete list of person_ids.
Input
| field | type | req | notes |
|---|---|---|---|
| segment_id | string | yes | Segment (UUID) |
| limit | number | no | default: 1000 (1–5000) |
Returns
segment_id — stringtype — string — manual, smart, or auto_streamperson_ids — Arraycount — numberconfig — Object (smart segments only)warnings — Array — predicate-evaluation warningslinks — Object (optional) — viewErrors
404 — Segment not found or not owned by youlist_segment_templatesfreeList available smart-segment templates. Pass a template id to create_segment to spin up a preconfigured segment.
Returns
templates — Array — id, name, default_segment_name, configmeta — Object — countsave_post1 creditSave a LinkedIn post to your library by URL. Resolves the post and stores it with source_type='saved'. Optionally adds the post to a manual collection. Failed lookups don't charge.
Save this LinkedIn post and drop it in my 'swipe file' collection.
get_alertsfreeReturn recent alerts for the authenticated user's streams.
Input
| field | type | req | notes |
|---|---|---|---|
| unread_only | boolean | no | default: true |
| limit | number | no | default: 50 (1–200) |
Returns
alerts — Array — id, stream_id, stream_name, items_new, message, read_at, created_at, linksmeta — Object — count, unread_onlylinks — Object — view_allErrors
400 — Query errorResearch (live LinkedIn lookups)
On-demand lookups against live LinkedIn data, metered in credits. Cached results come back free, paged tools charge only for pages fetched, and failed lookups are free. Post pulls land in your library, and per-post reactions/comments populate the engagement graph, so later questions about who engages with what are free.
linkedin_profile_info5 creditsResolve a LinkedIn person URL into a structured profile — full name, headline, current company/title, location, work history. Upserted into `people` and `linkedin_profiles` so downstream tools can filter on the fields without a re-pull.
Who is this LinkedIn profile?
linkedin_profile_posts5 credits / page of 50Fetch recent posts from a LinkedIn person profile, up to 250 (5 credits per page of 50, charged only for pages fetched). Results also stored in the post library (source_type='external') and linked to a linkedin_profiles row.
Pull recent posts from this LinkedIn profile.
linkedin_company_posts5 credits / page of 50Fetch recent posts from a LinkedIn company page, up to 250 (5 credits per page of 50, charged only for pages fetched). Stored in the post library.
linkedin_post_search5 creditsSearch LinkedIn posts by keyword. Returns a single page of matching posts.
linkedin_profile_reactions5 creditsFetch the recent posts a LinkedIn person has reacted to. Useful for inferring interests.
linkedin_profile_comments5 creditsFetch the recent posts a LinkedIn person has commented on. Useful for inferring interests and conversation partners.
linkedin_post_reactions1 credit / page of 100Fetch the people who reacted to a single LinkedIn post (with reaction kind), 100 per page, up to max_pages (default 10). Returns fetched / total_available / truncated. Engagers are persisted to the post_engagements graph.
linkedin_post_comments1 credit / page of 100Fetch the people who commented on a single LinkedIn post (with comment text and replies), 100 per page, up to max_pages (default 10). Each comment is persisted to post_engagements.
linkedin_people_search5 credits / page of 25Search LinkedIn people by company, title, location, or keywords, 25 per page. Returns each person's current title, company, company ID, and start date. Pass a company URL or numeric ID for an exact match. Results are saved so get_outreach_path can route to them.
linkedin_company_search5 credits / page of 50Search LinkedIn companies by keywords, industry ID, or location, 50 per page. Returns each company's LinkedIn ID.
linkedin_company_info1 creditFetch one LinkedIn company by URL or name: LinkedIn ID, industry, size, location, and website.
linkedin_job_search5 creditsSearch LinkedIn job postings. Useful for tracking hiring signals at target companies.
linkedin_ad_search10 creditsSearch the LinkedIn ad library. Useful for tracking competitor ad creative and messaging.
Constellation (warm outreach)
Map the graph around a target account, score warm paths, prioritize an outreach queue, and log outcomes so acceptance and reply rates build up over time.
create_campaignfreeCreate a Constellation outreach campaign. Groups target accounts, people you care about at those accounts, and warm connections you could approach.
Input
| field | type | req | notes |
|---|---|---|---|
| name | string | yes | 1–200 chars |
| description | string | no | max 2000 chars |
Returns
campaign — Object — id, name, description, status, created_atcredits_used — number — 0monthly_allocation — ObjectErrors
400 — Database insert failureadd_campaign_accountsfreeAdd one or more target companies to a campaign. map_constellation is then run per-account.
Input
| field | type | req | notes |
|---|---|---|---|
| campaign_id | string | yes | Target campaign (UUID) |
| accounts | array | yes | 1–500 per call; each: company_name, linkedin_company_id?, employee_count?, tier? |
Returns
campaign_id — string (UUID)added — number — inserted rowsaccounts — Array — id, company_name, linkedin_company_id, tier, status, created_atErrors
404 — Campaign not found or not owned by user400 — Database insert failuremap_constellationup to 1 + 3 × max_targetsMap the professional graph around a target company. Finds senior decision-makers, pulls each target's recent posts/reactions/comments to identify engagers, scores every connection, and produces a prioritized outreach queue. At most 16 credits at the default max_targets of 5 (61 at the maximum of 20); only what is spent is charged, and the response reports it. Returns the mapped targets by name, with next steps: the outreach queue, then a warm-path check for the top three targets.
Map the constellation for Acme Corp — who are the senior execs and who of mine is closest?
score_connectionsfreeRe-score all constellation entries in a campaign and rewrite queue positions from highest to lowest composite score. Run after new engagers or targets are added.
Input
| field | type | req | notes |
|---|---|---|---|
| campaign_id | string | yes | Campaign to re-score (UUID) |
Returns
campaign_id — string (UUID)rescored — number — entries re-scored and re-rankedErrors
404 — Campaign not found or not owned by userget_outreach_queuefreeReturn the prioritized outreach queue for a campaign, sorted by composite score. Each entry includes person details, pool (coworker/engager/mutual), recommended action, and prior log entries.
Who should I reach out to first this week?
Input
| field | type | req | notes |
|---|---|---|---|
| campaign_id | string | no | Full Constellation scoring (mutually exclusive) |
| segment_id | string | no | Engagement-count ranking (mutually exclusive) |
| linkedin_urls | array | no | Ad-hoc URLs, max 200 (mutually exclusive) |
| limit | number | no | default: 50 (1–200) |
| pool | enum | no | coworker, engager, mutual — campaign_id only |
Returns
scope — string — campaign, segment, or linkedin_urlsscoring — string — composite or engagement_countcount — numberqueue — Array — shape varies by scopeunresolved — Array (linkedin_urls mode only)Errors
400 — Must provide exactly one of campaign_id, segment_id, linkedin_urlsget_engagement_brieffreeFor a Constellation target, return the handful of their recent posts that most closely overlap with the topics you post about yourself — plus suggested talking points. Use before a warm comment or DM.
Brief me before I DM this person.
Input
| field | type | req | notes |
|---|---|---|---|
| target_id | string | no | Constellation targets.id (mutually exclusive) |
| person_id | string | no | people.id (mutually exclusive) |
| linkedin_url | string | no | Profile URL (mutually exclusive) |
Returns
scope — string — target, person, or linkedin_urltarget_id — string (UUID) | nulltarget — Object — full_name, headline, current_title, current_company, linkedin_urlcount — numberposts — Array — post_id, url, published_at, excerpt, engagement, similarity, talking_pointsErrors
400 — Must provide exactly one of target_id, person_id, linkedin_url404 — Target/person not foundlog_outcomefreeAppend an entry to the outreach log: connection requests, acceptances, replies, or no-responses. Feeds get_campaign_stats.
Input
| field | type | req | notes |
|---|---|---|---|
| campaign_id | string | yes | Campaign id (UUID) |
| person_id | string | yes | people.id (UUID) |
| action | enum | yes | sent_request, accepted, replied, no_response |
| note | string | no | max 4000 chars |
Returns
entry — Object — id, campaign_id, person_id, action, note, created_atErrors
404 — Campaign not found or not owned by user400 — Insert failureget_campaign_statsfreeSummary stats for a campaign: target count, entry count, outreach volume, and acceptance / reply rates broken down by pool and score tier.
Input
| field | type | req | notes |
|---|---|---|---|
| campaign_id | string | yes | Campaign id (UUID) |
Returns
campaign — Object — id, name, status, created_attotals — Object — accounts, targets, entries, outreach counts by statusacceptance_rate — number | nullreply_rate — number | nullby_pool — Object — per pool: sent, accepted, replied, ratesby_tier — Object — per high/medium/low: sent, accepted, replied, ratesErrors
404 — Campaign not found or not owned by userEngagement graph
Read-only queries against the engagement graph — people who reacted or commented on posts in your library. Mostly free; get_post_set_engagers fetches and enriches on demand. The graph is built up by stream runs and the paid linkedin_post_* tools, then mined by these.
get_post_engagers_by_companyfreeFor a post in your library, group its engagers by current company — counts plus optional sample names per company. Filter to a single industry URN to slice tighter.
Which companies showed up in the engagers on my last big post?
Input
| field | type | req | notes |
|---|---|---|---|
| post_url | string | no | Post URL (mutually exclusive with post_id) |
| post_id | string | no | posts.id (mutually exclusive with post_url) |
| industry_urn | string | no | Optional urn:li:fsd_industry:* filter |
| limit | number | no | default: 20 (1–100) |
| top_engagers_per_company | number | no | default: 3 (0–10) |
Returns
post_id — string (UUID)post_url — stringtotal_engagers — numberfiltered_engagers — numbercompanies — Array — company_id, name, industry_hierarchy, engager_count, top_engagers_meta — ObjectErrors
404 — Post not found in your library or tracked profiles400 — Must provide exactly one of post_url or post_idget_post_set_engagers1 credit / page + 3 / enrichConsolidated, deduped people who reacted/commented across a post set — one tracked profile's posts (profile_id) or a post collection (collection_id). Fetches only un-synced posts, 100 people per page (a typical post costs 2 credits); enriches only never-seen people up to max_enrich. Cached posts and pages are free.
Who engaged with the posts in my 'Q2 launch' collection?
enrich_people3 credits / person enrichedLooks up full LinkedIn profiles for up to 100 people and stores their title, company and job history, which get_outreach_path and get_alumni_engagers search. When those tools report missing job history, their _meta.next_step names this tool, the people to pass, and the cost. People you already have enriched in the last 90 days, failed lookups and not-found profiles are free.
Enrich the people get_outreach_path suggested, then find me a path to the target again.
get_engagement_industry_breakdownfreeHistogram of engagers' industries on a post, with optional depth (1 = top-level, 3 = leaf) and sample companies per bucket. Useful for sanity-checking ICP fit.
Input
| field | type | req | notes |
|---|---|---|---|
| post_url | string | no | Post URL (mutually exclusive) |
| post_id | string | no | posts.id (mutually exclusive) |
| depth | number | no | default: 1 (1–3; 1=top-level, 3=leaf) |
| sample_companies_per_bucket | number | no | default: 3 (0–10) |
Returns
post_id — string (UUID)post_url — stringtotal_engagers — numberdepth — numberindustries — Array — industry, industry_urn, engager_count, sample_companies_meta — ObjectErrors
404 — Post not found400 — Must provide exactly one of post_url or post_idget_alumni_engagersfreeFor a target company, list people who engaged with your posts and who are current employees, alumni, or both (configurable). Returns title-at-company, current role, and engagement count.
Who at Acme has engaged with my content?
Input
| field | type | req | notes |
|---|---|---|---|
| company_url | string | no | Company URL (mutually exclusive) |
| company_id | string | no | companies.id (mutually exclusive) |
| linkedin_company_id | string | no | Numeric LinkedIn company ID (mutually exclusive) |
| company_name | string | no | Exact name, ignoring case; ambiguous names return candidates (mutually exclusive) |
| relation | enum | no | default: any (coworker, alumni, any) |
| limit | number | no | default: 20 (1–100) |
Returns
company — Object — id, name, linkedin_url, page_type, industry_hierarchyrelation — stringtotal_engagers — numberengagers — Array — person details + title_at_company, is_current_at_company, engagement_countcoverage — Object — candidates considered, with and without job history_meta — Object — interpretation; next_step names enrich_people, the people to pass, and the cost when missing job history may hide matchesErrors
404 — Company not foundresolve_peoplefreeLook up to 200 LinkedIn URLs or person IDs in your graph at once: who each one is, whether job history is on file, how they connect to you, and the enrich_people call that fills the gaps. People not connected to you come back the same as unknown URLs.
Which of these 40 profiles does Flux already know, and what would it cost to fill in the rest?
Input
| field | type | req | notes |
|---|---|---|---|
| linkedin_urls | string[] | no | Profile URLs, any form (200 total with person_ids) |
| person_ids | string[] | no | people.id UUIDs |
Returns
results — Array — input, person_id, linkedin_url, matched_via (exact | canonical | member_id | public_identifier | none), full_name, current_title, current_company, has_positions, enriched_at, engaged_with_you, in_campaigns, connected_viasummary — Object — requested, matched, unmatched, with_job_history, engaged_with_you_meta — Object — interpretation; next_step = enrich_people with the unmatched URLs and the people missing job history, plus its costErrors
400 — Provide linkedin_urls and/or person_ids; at most 200 inputsresolve_companyfreeLook up one saved company by name, LinkedIn URL, LinkedIn numeric ID or Flux ID, with how many people connected to you work there now and how many of them engage with your posts.
Input
| field | type | req | notes |
|---|---|---|---|
| name | string | no | Exact, case-insensitive (exactly one of the four) |
| linkedin_url | string | no | linkedin.com/company/… |
| linkedin_company_id | string | no | LinkedIn numeric ID |
| company_id | string | no | companies.id |
Returns
company — Object — company_id, linkedin_company_id, name, linkedin_url, page_typecoverage — Object — connected_people_current, engagers_current, engagements_current, truncatedcandidates — Array — when several saved companies share the name_meta — Object — next_step = linkedin_company_info (1 credit) when the company is not savedErrors
404 — Company not found (with next_step)400 — Generic placeholder nameget_engager_position_historyfreeCareer history of a person who has engaged with your posts. Engagement gate: only available for people who actually engaged (not arbitrary LinkedIn profiles).
Input
| field | type | req | notes |
|---|---|---|---|
| person_id | string | no | people.id (mutually exclusive) |
| linkedin_url | string | no | Profile URL (mutually exclusive) |
Returns
person — Object — id, name, linkedin_url, headline, current_company, current_title, locationpositions — Array — title, company_name, company_id, is_current, started_at, ended_at, source_tool_meta — ObjectErrors
404 — Person not found400 — Person has not engaged on your posts (engagement gate)get_target_warmth_brieffreeFor an engager, return their engagement summary against your posts: warmth score (0–1.0, v0), reaction/comment counts, first and most recent engagement dates, and the actual posts they touched. Use to qualify a warm DM.
How warm is this person before I reach out?
Input
| field | type | req | notes |
|---|---|---|---|
| person_id | string | no | people.id (mutually exclusive) |
| linkedin_url | string | no | Profile URL (mutually exclusive) |
| days_back | number | no | default: 90 (1–365) |
Returns
target — Object — id, name, headline, linkedin_url, current_company, current_title, locationwarmth_score — number — 0–1.0 (v0)engagement_count — Object — reactions, comments, totalfirst_engagement — string (ISO)most_recent_engagement — string (ISO)posts_engaged — Array — per-post engagement details_meta — ObjectErrors
404 — Person not foundget_company_engagement_overlapfreeCross-company intel: people who have engaged with both Company A's and Company B's own company-page posts in your library (their audiences, not their employees). Returns engagement counts per company and totals. Ambiguous names return candidates; an unknown company suggests linkedin_company_info.
Input
| field | type | req | notes |
|---|---|---|---|
| company_a_id / company_a_url | string | no | Company A by Flux id or URL (exactly one identifier per company) |
| company_a_linkedin_id / company_a_name | string | no | Company A by numeric LinkedIn id or exact name (ignoring case) |
| company_b_id / company_b_url | string | no | Company B by Flux id or URL |
| company_b_linkedin_id / company_b_name | string | no | Company B by numeric LinkedIn id or exact name |
| limit | number | no | default: 20 (1–100) |
Returns
company_a — Object — id, name, linkedin_url, page_type, industry_hierarchycompany_b — Object — same shapeoverlap_count — numberengagers_on_a_only — numberengagers_on_b_only — numberengagers — Array — per-person engagement counts to each company_meta — ObjectErrors
404 — Company A or B not found400 — Must provide exactly one identifier for each companyget_outreach_pathfreeFind warm-intro paths to a target person via coworker/alumni edges in your engagement graph. Configurable max hops (1–3). Returns ranked paths grouped by hop count, and says whether an empty result means no connection or not enough data.
Who's my warmest path to this VP?
Input
| field | type | req | notes |
|---|---|---|---|
| target_person_id | string | no | Target id (mutually exclusive) |
| target_linkedin_url | string | no | Target URL (mutually exclusive) |
| max_hops | number | no | default: 2 (1–3) |
| limit | number | no | default: 10 (1–100 paths) |
Returns
target — Object — id, name, linkedin_url, current_company, current_titlemax_hops — numberpaths — Array — warm_engager + steps + hopspaths_by_hop_count — Object — keyed by hop counttarget_engaged_directly — the target's own engagement with your postscoverage — Object — whether the target and the engagers tied to their companies have job history_meta — Object — interpretation (no connection vs not enough data); next_step names enrich_people, the people to pass, and the costErrors
404 — Target person not foundGraph traversal
experimentalRun read-only SQL queries against a sandboxed view of your data. Useful for exploration when the higher-level tools don't fit. v0: validation is regex-based (substring match on caller IDs, keyword filtering); real RLS lands later. Treat as experimental and expect rough edges.
traverse_graph_helpfreeReturns the table schema, FK edges, your caller identifiers (caller_user_ids and tracked_lp_ids that traverse_graph requires you to reference), example queries, and the validation constraints. Call this first.
Show me the graph schema and an example query.
Returns
tables — Object — schema for posts, post_engagements, people, person_positions, companies, linkedin_profiles, tracked_profilesfk_edges — Array[string] — FK relationshipsexamples — Array — 5 canonical SQL examplescaller_user_ids — Array[string] — your user_ids (email-shared)caller_tracked_lp_ids — Array[string] — your tracked LinkedIn profile idsrls_notes — stringconstraints — Object — forbidden_keywords, must_start_with, max_rows, statement_timeout_seconds_meta — Objecttraverse_graphfreeExecute a single SELECT (or WITH ... SELECT) against the graph. Must reference one of your caller_user_ids or tracked_lp_ids — otherwise rejected. INSERT / UPDATE / DELETE / DDL are blocked. Row cap 1000; statement timeout enforced.
Run a custom query: people who engaged with at least 3 of my posts in the last 60 days.
Input
| field | type | req | notes |
|---|---|---|---|
| sql | string | yes | SELECT or WITH; must reference caller_user_id or tracked LP id; min 10 chars |
Returns
rows — Array — result rows from queryrow_count — numbergenerated_sql — string — echoed backelapsed_ms — number_meta — Object_warnings — Array — present if row count capped at 1000Errors
400 — SQL does not start with SELECT or WITH400 — SQL contains forbidden keywords (INSERT, UPDATE, DELETE, etc.)400 — SQL does not reference caller_user_id or tracked LP id400 — Query execution errorAdmin & billing
Visibility into credit balance, tool pricing, and recent usage.
get_credit_balancefreeReturn the authenticated user's current credit balance, this period's monthly allocation, and credits used since the period started.
How many credits do I have left?
Returns
balance — numbermonthly_allocation — Object — amount, plan_type, period_start, period_end (or null)used_this_period — numberget_cost_estimatefreeGiven an MCP tool name or a stream type, return its credit cost and whether your balance covers it, so you can preview an action before running it. Pass the planned arguments in params to get the worst case for that exact call, including paged tools.
What will map_constellation cost me on Acme?
Input
| field | type | req | notes |
|---|---|---|---|
| tool_name | string | yes | MCP tool name or stream type |
| params | object | no | The arguments you plan to pass; paged tools are priced at their worst case (e.g. get_post_set_engagers, 6 posts at defaults → 120) |
Returns
tool_name — stringcost_max — number — the most the call can spendcost_typical — numberunit — string — e.g. 1 per pageexplanation — stringcost — number — equals cost_max (0 if not in registry)balance — numbersufficient — boolean — balance >= costknown_tool — booleanstream_cost — Object — credits_per_page, posts_per_page, max_pages_ceiling (stream types only)get_usage_historyfreeRecent credit_ledger transactions for the authenticated user. Filter by tool, reason, or date range. Paginated.
Input
| field | type | req | notes |
|---|---|---|---|
| tool_name | string | no | Filter to a tool |
| reason | enum | no | monthly_grant, pack_purchase, tool_usage, refund, adjustment |
| date_from | string | no | ISO 8601 start |
| date_to | string | no | ISO 8601 end |
| limit | number | no | default: 50 (1–200) |
Returns
transactions — Array — id, amount, balance_after, reason, tool_name, stream_id, metadata, created_atmeta — Object — count, filters_appliedErrors
400 — Query errorWho it’s for
FAQ
Edge is $15/month. Cancel anytime. Connect Claude the moment you subscribe.