Model Context Protocol

The Justify MCP server
for AI agents.

Give your agents direct access to Justify — the measurement-first platform for influencer marketing. Search 300M+ creators with evidence, run vetting pipelines, and (soon) prove what moved, through a single MCP connection.

Get started
Available on Essential plan and above
Terminal 
$ claude mcp add --transport http justify https://mcp.justify.app/mcp
Influencer Marketplace · live now · Creative Testing · coming soon · Brand Lift · coming soon
01 · What you can build

Connect once.
Your agent does the rest.

One MCP connection covers the whole platform. The Influencer Marketplace is wired up today — Creative Testing and Brand Lift land on the same connection as they ship.

● Live
Influencer Marketplace

Discover & vet creators at scale

Search 300M+ creators across Instagram, TikTok, and YouTube. Build personas. Run sourcing pipelines that vet every creator against your brand criteria — agents review and approve, you ingest the winners.

→ Full tool surface
Coming soon
Creative Testing

Proof before the post goes live

Pull live test results, rank concepts by lift, and feed the winner back into the brief. Same MCP connection — new tools appear automatically when the surface ships.

→ Same connection
Coming soon
Brand Lift

Read studies, decide next move

Pull exposed-vs-control reads with confidence ranges, query past studies for benchmarks, and trigger new pre-registered designs from agent prompts.

→ Same connection
02 · One connection, three surfaces

The whole platform, one connection.

No per-surface APIs to integrate. The MCP server presents the whole platform behind a single endpoint — the tools your agent sees expand as new surfaces ship, no client changes required.

Your client Claude · Cursor · custom agent
MCP · OAuth mcp.justify.app/mcp
Live Influencer Marketplace
Coming soon Creative Testing
Coming soon Brand Lift
03 · Quick setup

Connect in two steps.

Works with Claude Code, Claude Desktop, Cursor, and any client that speaks MCP over HTTP. Authenticate via OAuth on first use — no API keys to copy or rotate.

  1. 1

    Sign up or log in

    Sign up to Justify, then make sure you're on Essential or above. Access is granted via OAuth — no API keys required.

    Go to Justify
  2. 2

    Add to your MCP client

    Paste the config into Claude Code, Claude Desktop, Cursor, or any MCP-compatible client. You'll be prompted to authenticate via OAuth on first use. Tools appear automatically.

Terminal 
$ claude mcp add --transport http justify https://mcp.justify.app/mcp
Restart Claude Code, run /mcp to verify.
04 · Available tools

A full toolset, one connection.

Every tool your agent needs to discover influencers, manage personas, run sourcing pipelines, and (soon) prove what moved. Tool names are stable — breaking changes ship behind a versioned namespace.

Run autonomous sourcing and vetting pipelines against a persona. Review results, approve or discard runs, and override individual vetting outcomes.

initiate_sourcing_run
Start a sourcing and vetting pipeline for a persona. Sources influencers matching the persona, then vets each one against the criteria.
list_sourcing_runs
List sourcing run history for a persona with status and influencer counts per run.
get_sourcing_results
Get the status and results of a sourcing run, including pass/fail outcomes with detailed reasoning.
approve_sourcing_run
Approve an on-hold sourcing run and ingest the passed influencers into the target campaigns.
discard_sourcing_run
Discard an on-hold sourcing run without ingesting any influencers.
override_vetting_result
Override the vetting result for a specific influencer — approve one who failed or reject one who passed — before approving a run.
export_sourcing_results_csv
Export sourcing results for a pipeline run to a CSV file.
get_sourcing_usage
Get the brand's sourcing usage and entitlement limits for the current billing period.
05 · Example workflows

What your agent actually does.

Once connected, your agent runs the sourcing-and-vetting loop without you babysitting it. One live workflow today. Two more land as the surfaces ship — same connection, no client changes.

● Live
01 — Sourcing & vetting

Discover and vet brand-aligned creators at scale

Persona-led pipeline that sources, vets, and presents pass/fail results for review.

  1. 1Agent calls create_persona with your ideal influencer profile, demographics, and vetting questions
  2. 2Agent calls initiate_sourcing_run to source and vet influencers matching the persona at scale
  3. 3Agent calls get_sourcing_results to review pass/fail outcomes, then approve_sourcing_run to ingest
Coming soon
02 — Creative testing

Rank concepts before spend

Read live test results, surface winners, and feed them back into the brief.

  1. 1Agent pulls test results across active concepts
  2. 2Agent ranks concepts by predicted lift and confidence
  3. 3Agent updates the brief with the winning frame and recommended cell sizes
Coming soon
03 — Brand lift

Read studies, decide next move

Pull pre-registered exposed-vs-control reads with confidence ranges and benchmarks.

  1. 1Agent retrieves the latest study with confidence range and p-value
  2. 2Agent compares against historical benchmarks for the category
  3. 3Agent compiles a one-page read with the recommended next action
06 · FAQ

Common questions.

WhatWhat is MCP?+

MCP — Model Context Protocol — is an open standard that lets AI agents connect directly to external services. Instead of manual data pulls or custom integrations, your agent can call Justify's tools natively — searching creators, building personas, vetting influencers, and triggering pipelines in real time, all within the same conversation.

ClientsWhich MCP clients are supported?+

Claude Code (CLI), Claude Desktop, Cursor, and any client that supports the MCP HTTP transport. Add the server URL and authenticate via OAuth — tools appear automatically.

AuthHow does authentication work?+

OAuth, in your browser. When you first connect, you'll be prompted to sign in to Justify and grant the client access. Sessions are workspace-scoped and tokens are managed securely by your MCP client — no API keys to copy, paste, or rotate.

PlansWhich plans include MCP access?+

Essential and above. Starter does not include MCP — the connection requires the workspace, persona, and sourcing infrastructure that begins on Essential. Every paid tier above gets the same tool surface; plan limits gate volume, not capability. See pricing for what's included on each tier.

LimitsWhat are the rate limits?+

MCP usage draws from your plan's monthly credit allowance, with unused credits rolling forward. Burst rate limits keep traffic stable for everyone; if you hit one, you'll get a 429 with a Retry-After value in seconds. Need more headroom? Talk to us.

DataHow is my data handled?+

OAuth scopes are explicit: your client can only see what your role can already see in the Justify app. We don't train models on your data. Justify's own models are trained on public creator content and licensed datasets — nothing else. Every tool call is recorded in your workspace audit log: who, what, when, with what arguments. The same security posture as the rest of the app — see /security for the full picture.

VersionsHow are breaking changes handled?+

Tool names follow semver. Additive changes — new tools, new optional fields — ship without notice. Breaking changes ship behind a versioned namespace (v2.search_influencers); the old name keeps working through a deprecation window and the change appears in the audit log so you can plan the migration.

Connect your agent.
Justify the spend.

Sign up