MCP for ChatGPT, Claude, Codex, and compatible clients

Evaluate CSV rowswithout filling your assistant's context.

Send per-record questions to Decision Cloud through MCP. Get structured judgments back in your assistant, where it can interpret and rank the results.

  • ChatGPT
  • Claude
  • Codex
  • Other MCP clients

In your assistant

Ask what to check

YouYour request

For each row in this CSV, assess buying intent and product fit.

CSV file attached
The host sets the questions

MCP service

Decision Cloud

Ready
TOOL CALLevaluate_file
InputCSV rows
Per-row questions
IntentFit
Row limits200 by default · 500 max

One structured judgment for each selected row

Back in your assistant

Review the results

Synthetic example

Decision Cloud returns a judgment for each row. Your assistant can compare the results and explain what to do next.

RowIntentFit
Record 01LikelyHigh
Record 02UnclearMedium
Record 03PossibleReview

No context bloat

Keep the CSV out of the host's working context. The host sends questions and gets judgments back.

No Jev markup

Jev compute passes through at its configured rate. Decision Cloud adds no compute markup.

Results back in your agent

Structured judgments return to the host, which compares and ranks the rows.

Give your assistant room to think about the answer.

A file can bring useful context, but sending every row through the host can crowd out the request. Delegate the repeated checks and keep interpretation with your assistant.

Your host

Frames the question

Choose the context and per-row judgments that fit the task.

Decision Cloud

Runs each row's checks

Return structured judgments without asking the host to carry every row.

Your host

Interprets and ranks

Compare the returned judgments, decide what matters, and explain the result.

Bring the rows that need a judgment.

Decision Cloud evaluates your assistant's questions one row at a time. Your assistant decides what the returned results mean.

Sales lead prioritization

Ask for buying intent and product fit on each lead. Let your assistant compare those judgments and explain where to start.

Questions your host can ask
Buying intentProduct fitFollow-up need

Resume and job matching

Compare each resume with a role's requirements. Your assistant can explain which experience lines up with the job.

Support ticket triage

Label tickets by issue or urgency. Your assistant can group the results and suggest what to address first.

Pay for what you use.

Add credits when you need them. There's no subscription and no Decision Cloud markup on Jev compute.

See pricing details
Decision Cloud creditsChoose an amount
Service fee per top-up$1.00
SubscriptionNone

Jev compute is charged at its configured rate, with no markup.

Connect in minutes.

Your host remains the place where you describe the task and read the results.

  1. 1

    Create an account

    Sign in to manage your Decision Cloud account.

  2. 2

    Add prepaid credits

    Top up when you need more evaluation balance.

  3. 3

    Connect the MCP server

    Follow the setup guide for your assistant.

  4. 4

    Ask for the checks you need

    Your assistant sends questions and reviews the results.

Keep the rows out. Bring the answers back.

Let your assistant define the checks, then use Decision Cloud to run them across the CSV.