Linkfetch
get key ↗
sheet 00 / 08—briefing / daily outreach
the daily outreach agent

The LinkedIn
interface
for your
AI agents.

One typed call gives Claude, Cursor, and any MCP client read access to profiles, companies, and jobs — through your own signed-in LinkedIn session. No scraping, no fake accounts.

run this for my signupsread the recipe
invites / day
10
accept rate
70%
hands-on
5 min

user is principal ● empty results never charged ● public data only ● gdpr + ccpa aligned

instrumentlinkfetch.autopilotlive
promptsaved · 2026-04-18·runs daily · 09:00

“Every morning, find 10 new VPs of Engineering hired at Series-B fintechs. Draft a 2-line warm intro that names their last role — then queue the connection requests through my LinkedIn.”

complete·search·enrich·company·compose·queue·on_accept
100%
run log200thu · 07:13
mcp · linkfetch + claude
[07:03]linkfetch.people.search10 matches · 0.6s[07:03]linkfetch.profiles.get ×1010/10 enriched · 50 credits[07:03]linkfetch.companies.get ×1010/10 · 80 credits[07:04]claude.compose intros10 drafts · 3.8s[07:04]linkfetch.outreach.queue10 queued · user is principal[07:12]on_accept → follow_up7 accepted · 5 drafting…
prospects10today's run
accepts770% · on track
credits130≈ $0.65 · daily
ll-0077 · daily-outreachrecipe v1 ●
also runs well fornew head of data · saasvp sales · b2b devtoolspm hires · ai-nativefounding eng · yc w26rev-leadership · seed+recent leavers · stripe
starter briefs · §00b

Tell Claude something real.
Six starters, ready to ship.

Every brief below is a real prompt our users run. Copy it into Claude, swap the account, the role, or the cadence — ship it before lunch.

01/hiring signals
one-shot
ask claude
“Who are the 5 most recent hires at Vercel, and what were they doing right before?”
linkfetch.profiles›linkfetch.companies
~20 creditscopy brief →
02/churn watch
weekly
ask claude
“Who left Stripe in the last 60 days, and where did they land? Flag anyone who joined a direct competitor.”
linkfetch.profiles.search
~60 credits / weekcopy brief →
03/daily outreach
daily · 09:00
ask claude
“Every morning, find 10 new VP Engineering hires at Series-B fintechs and queue a 2-line warm intro for each.”
linkfetch.people›linkfetch.profiles›linkfetch.companies›claude.compose›linkfetch.outreach.queue
~130 credits / daycopy brief →
04/competitive intel
daily
ask claude
“Track headcount at Anthropic, OpenAI, and Mistral. Ping me on ≥5% swings or notable new hires.”
linkfetch.companies.timeseries›linkfetch.profiles
~8 credits / daycopy brief →
05/company brief
on demand
ask claude
“For linear.app: size, last funding, headcount trend, and 3 notable new hires this quarter.”
linkfetch.companies›linkfetch.profiles
~40 creditscopy brief →
06/icp enrichment
per signup
ask claude
“For every new trial user: role, seniority, company size, funding stage. Rank by ICP fit and score.”
linkfetch.profiles›linkfetch.companies
~13 credits / usercopy brief →

Every brief runs inside Claude Desktop, Cursor, or anything that speaks MCP. No custom glue — just the LinkFetch Skill and your API key.

install mcp server ↗all 40+ recipes
endpoints · §02

Three data surfaces.
One compliance story.

Every endpoint returns typed JSON with a provenance stamp and a predictable credit cost. Public data only, aligned with GDPR and CCPA — your users stay the principal on their own session.

01/profiles
5credits
APERTUREf/1.0 · fullfocused

The person, the whole résumé.

Name, headline, roles, education, skills — fetched through the user's own LinkedIn session via the LinkFetch Chrome extension. No fake accounts, no session impersonation; the user is the principal.

GET /v1/profilesdocs →
06/http
—credits
python · requests03
curl · sh02
typescript · tstyped
const res = await fetch(
  "https://api.linkfetch.io/v1/profiles?url=" + url,
  { headers: { Authorization: "Bearer " + key } },
);
const { data, meta } = await res.json();
ts · py · sh · anyone envelope · every runtime

Plain HTTPS, typed JSON, any language.

No SDK to install: one bearer header and a JSON envelope that is identical on every endpoint. Consistent error codes and a request id on every response so support tickets write themselves.

fetch · requests · curldocs →
02/companies
8credits
headcount · 28q
+24% ▲

Size, funding, headcount trend.

Resolve slug → structured company record. Ships with last-funding and employee-growth signals when available.

GET /v1/companies/:slugdocs →
03/jobs
3credits
new roles · 12 · last 24h
SF · NYC · LON

Fresh listings with first-seen.

Jobs by keyword, company, seniority — first-seen timestamps make this the signal, not just the data.

GET /v1/jobsdocs →
04/provenance
—credits
verified
fetched 2026-04-23
source ● extension
meta.provenance

Every row has a source stamp.

source, fetched_at, freshness_days on every response. Auditable, explainable, agent-safe.

meta.provenancedocs →
05/compliance
—credits
GDPRCCPA
public data only

Aligned with GDPR and CCPA.

Public data only, no fake accounts, no session impersonation. Your users stay the principal on their own LinkedIn session, with a clear legal story for EU and California workloads.

GDPR · CCPAdocs →
§03 · claude-native

Ask Claude “who works at Stripe?”
— get a real answer.

Add the MCP server to Claude Desktop or Claude Code with one config block. LinkFetch shows up as native tools; your Claude conversations get LinkedIn context the moment they need it.

install mcp server ↗claude code setup
claude_desktop_config.jsoncopy ⧉
{
  "mcpServers": {
    "linkfetch": {
      "command": "npx",
      "args": ["-y", "linkfetch-mcp"],
      "env": { "LINKFETCH_API_KEY": "sk_live_…" }
    }
  }
}
C
Claudelinkfetch mcp · connected
online
you · 09:41
Who are the 5 most recent hires at Vercel?
tool call
linkfetch.search_people { company: "vercel", sort: "joined_desc", n: 5 }
1.2s · 15 credits · 3 profiles
claude · 09:41
Here are the 5 most recent hires I can see:
  1. 01Priya Shah— Staff Engineer · mar 2026
  2. 02Elena Ruiz— PMM, DX · mar 2026
  3. 03Tomás Vieira— Backend, Fluid compute
  4. 04Jordan Lee— Eng Manager, edge
  5. 05Aida Khan— SDR, EMEA
stream idle
§04 · closing

Ship LinkedIn context
before the week ends.

$5 in credit, no card, public data only, provenance on every row. Wire it into Claude, Cursor, or your own agent stack — and keep shipping on Friday.

get api key — $5 free ↗read the docs

no card ● empty results never charged ● public data only ● gdpr + ccpa aligned

§05 · quickstart

Three steps from zero
to production.

Paste a key, ship a call. No SDK to install, no waiting room, no annual commit, no sales call — if you have a terminal and 90 seconds you have an API.

live sandbox/$5 free credit on sign-up
01/key
step 01
export LINKFETCH_KEY=sk_live_…

curl "https://api.linkfetch.io/v1/jobs?q=staff+engineer&posted_within=week" \
  -H "Authorization: Bearer $LINKFETCH_KEY"

Bring a key. Nothing to install.

Sign in, copy the key from the dashboard. Scoped, revocable, rate-limited per tier. No fake accounts, no session tokens stored — the API is plain HTTPS and JSON.

1 credit · empty rows freedocs →
02/call
step 02
const res = await fetch(
  "https://api.linkfetch.io/v1/profiles?url=linkedin.com/in/jane-doe",
  { headers: { Authorization: `Bearer ${process.env.LINKFETCH_KEY}` } },
);
const { data, meta } = await res.json();
// data.headline, meta.provenance

Typed JSON, provenance stamped.

Every response ships with source, fetched_at, and freshness_days in meta. Profiles and companies resolve through the user's own session via the extension.

5 credits · fetch or any http clientdocs →
03/agent
step 03
{
  "mcpServers": {
    "linkfetch": {
      "command": "npx",
      "args": ["-y", "linkfetch-mcp"],
      "env": { "LINKFETCH_API_KEY": "sk_live_…" }
    }
  }
}

Or let the agent call it.

The MCP server exposes the same endpoints as native tools for Claude Desktop, Claude Code, Cursor, and Zed. One config block, no fetch wrapper.

claude · cursor · zed · continuedocs →
credits per jobs search
1
first successful call
~90s
annual commit
none
empty results charged
0
§06 · how we compare

Built for 2026 —
not 2018.

The market opening when Proxycurl shut down in July 2025 made one thing clear: the old compliance story doesn't hold. LinkFetch is the drop-in replacement with a legal story that survives.

capabilityLinkFetchCoresignalBright DataProxycurl(shut down Jul '25)
01Time to first call
90 secDaysHours—
02Free tier
$5 creditTrialno—
03Annual commit required
noyesyes—
04Claude / MCP native
yesnono—
05Private profiles via user's own extension
yesnonono
06GDPR + CCPA aligned
yesPartialPartialno
07Provenance on every row
yesnonono

ref. ll-0044 · observations current as of rev.2026.04 · public docs + pricing pages

§07 · ledger

Pay for what you use.
Empty rows are free.

Credits with a transparent per-query table. No annual commit, no minimums, no surprise seat fees — and if a call returns nothing, we don't charge.

freetier
$0$5 credit · no card

Kick the tires. Enough for a full weekend prototype.

  • ▸Thousands of jobs calls
  • ▸1 req/sec
  • ▸Email support
  • ▸All endpoints
Get API key →
most picked
startertier
$29/month

Solo builders shipping a side project to a handful of users.

  • ▸~6,000 lookups at avg mix
  • ▸10 req/sec
  • ▸GDPR + CCPA aligned
  • ▸Email support
Start Starter →
protier
$99/month

Funded startups with production agent workloads.

  • ▸~25,000 lookups
  • ▸50 req/sec
  • ▸Priority support
  • ▸GDPR + CCPA aligned
Start Pro →
scaletier
Customlet's talk

For teams with dedicated volume and compliance asks.

  • ▸Custom throughput
  • ▸Private MCP cluster
  • ▸DPA, SSO, invoice
Talk to us →
credits per requestrev.2026.04
requestendpointcredits
01profile lookup
GET /v1/profiles5
02company lookup
GET /v1/companies/:slug3
03jobs search
GET /v1/jobs1
04empty result
404 / 0 rows0
§08 · datasets

Or just buy the snapshot.

Prefer bulk over API calls? The whole LinkedIn jobs archive — 8M+ public postings since April 2026 — as a PostgreSQL dump, CSV, and JSONL, built fresh at purchase. Same provenance, same compliance, no credits.

01/jobs · linkedin
fresh · last 24h
new roles · worldwide
45,000+
indexed per day/218 regions
total postings8M+
archive sinceApril 2026
applicant historyper posting
jobs dataset · $199 one-time ↗see sample
postgres · csv · jsonl · or $49/mo live

public data only ● no fake accounts ● gdpr + ccpa aligned ● dpa on request for scale snapshots

§ faq

Straight answers, no sales call.

  • What is LinkFetch?

    A LinkedIn data API for AI agents. It returns public LinkedIn profiles, companies and job postings as typed JSON over plain HTTPS, exposes the same surface as MCP tools for Claude, Cursor and Zed, and sells its 8M+-row jobs index as a flat-price dataset. It is not affiliated with LinkedIn.

  • Is it a Proxycurl alternative?

    Yes, and a differently built one. Proxycurl ran rented accounts and shut down in July 2025. LinkFetch reads jobs from LinkedIn's public guest endpoints and reads profiles and companies through the end user's own session via a Chrome extension, so the user is always the principal. Field names overlap about 90 percent; a migration guide and field map are on the blog.

  • How much does it cost?

    Flat credits per request: a profile lookup is 5 credits, a company lookup 3, a jobs search 1. Empty results are free. The free tier includes $5 of credit with no card; Starter is $29/month and Pro $99/month. The jobs dataset is $199 one-time or $49/month for live access.

  • Do I need an SDK?

    No. The API is plain HTTPS and JSON, so fetch, requests or curl work as-is. Agents can use the MCP server (linkfetch-mcp) instead and skip HTTP entirely.

  • Is the data legal to use?

    Jobs are public postings any logged-out visitor can see. Profiles and companies are fetched in the user's own session, never with fake or rented accounts. Every response carries a provenance stamp with source, fetched_at and freshness_days, and data subject requests are honoured with suppression by default.

  • How fresh is the jobs data?

    The index adds about 45,000+ postings per day across 218 regions and re-checks is_active on recrawl. Each row carries first_seen_at and last_seen_at, and postings with applicant counts get a timestamped applicant-count history.