Somebodyowes youmoney.

Class-action settlements only pay the people who file. Most never do — not because they were ineligible, but because nobody told them a company they used, worked for, or bought from ever settled.

Your AI already knows the clues. Redra hands it the whole docket. ChatGPT, Claude, and any MCP-compatible agent can search it against the life they already know: the brands, the subscriptions, the employers, the city, the years.

Connect freeRead the source

Free hosted MCP · No account · No prompts or model memory stored

redra / docketLIVE
Model memory

You used Amazon Prime, tracked your period with Flo, and live in California.

2 real leads returned
  1. Match

    Amazon Prime Membership FTC Settlement

    Open for claims · Deadline Jul 27, 2026

  2. Match

    Flo Health App $59.5M Privacy Class Action Settlement

    Open for claims · Deadline Oct 15, 2026

2 real leads worth verifyingVerify before filing
$179.6Bup for grabs
367open for claims
Dailydataset refresh
Zeroprompts or memory stored

Bring the agent you already use

Supported platforms: ChatGPT, OpenAI Codex, Claude.ai, Claude Desktop, Claude Code, Cursor, VS Code, GitHub Copilot, Gemini CLI, Windsurf, Cline, Continue, Zed, JetBrains AI, Kiro, goose, LM Studio, LibreChat, Copilot Studio, Custom MCP client.

§ 01Recall

You don’t need to know what to search.

Ask broadly. Your agent turns the messy details it already knows into precise Redra searches, then brings back the leads worth verifying.

@Redra Check the companies, products, and subscriptions I’ve used for anything I might qualify for.

@Redra Use my connected receipts and transaction history to identify possible settlement matches.

@Redra Dig through what you remember about me and search from every reasonable angle.

Invocation varies by client. Connect Redra, then ask naturally.

What your agent knowsWhat Redra searches
Amazon Prime since 2016subscriptions · auto-renewal · junk fees
Flo, for cycle trackingprivacy · data sharing · health apps
Chase checking accountoverdraft · NSF · bank fees
Lives in Los Angeles, CACalifornia class · nationwide class · resident claims
Born 1994, household of fourage eligibility · household size · family status
Nurse, 2019 through 2023occupation · employers · wage claims

Six lines of context. Eighteen searches worth running.

§ 02Method

Your memory, meet the right database.

  1. 01

    Bring what you already know

    Your agent holds the brands, subscriptions, employers, purchases, and places that quietly decide eligibility. None of it has to be typed into a search box again.

    Context in
  2. 02

    Search from every angle

    Redra turns that context into focused, structured searches across a settlement dataset that refreshes daily — parent companies and aliases included.

    Recall out
  3. 03

    Verify, then claim

    Each hit comes back as a lead: deadline, proof requirement, potential payout, and the official filing link. A lead, never a legal conclusion.

    Human decides

§ 03In tandem

You are limited only by your agent.

Pair Redra with the plugins and connectors your agent already has — email receipts, bank transactions, subscriptions, purchase history, settlement notices. Your agent converts that into non-sensitive brands, products, and incidents. Redra searches those.

§ 04Filters

Broad recall. Exact filters.

Start creatively, then narrow with structured fields a model can use reliably. Every result keeps the source details a human needs to verify it.

FieldAcceptsNotes
KeywordsCompanies · products · incidentsCombine precise terms with logical AND, or run separate searches for broader recall.
Claim statusOpen · closed · payment · allOpen claims are the default, with the full lifecycle available when you need it.
Settlement typePrivacy · breach · fees · refundsStructured categories, instead of hoping the type happens to appear in a title.
LocationState or nationwideFilter by a two-letter US state while retaining settlements open nationwide.
Claim detailsProof · deadline window · payoutNarrow by proof requirements and filing dates, then review potential compensation.
Source detailsFiling links · last checked · warningsWhere the result came from, when it was checked, whether a claim link exists, and any warnings.

Results can include filing links, estimated payout, claim deadline, applicable states, proof requirements, when the source was last checked, and important warnings. Up to 50 results per search.

§ 05Deploy

Privacy-oriented. Fast. Yours.

Hosted Redra

Recommended

Point your client at the public MCP endpoint and search the maintained dataset for free. No server, no database, no API key.

  • Daily settlement updates
  • No submitted data, prompts, or model memory stored
  • Capacity safeguards for shared AI clients
  • Works with any MCP client
Use the hosted MCP

Self-hosted Redra

Open source

Run the MCP server and dataset in your own environment. You own the infrastructure, the refresh schedule, the access policy, and the privacy boundary.

  • Docker-ready deployment
  • Keep the database and search service on your network
  • Choose your own access, retention, and refresh policy
  • Extend or modify every part of the server
Read the self-host guide

§ 06Connect

Connect Redra. Free.

Add the Redra MCP endpoint to ChatGPT, Claude, or any other MCP client for free. Redra does not collect or store submitted data, prompts, or model memory.

MCP endpointhttps://mcp.redra.ai/mcp

Streamable HTTP · no key required