Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Ask verified humans

    1

    Let your agent check judgment calls with ID verified people before it acts. Built on plures-mcp.

    Free

    Security scanned
    Ask verified humans

    Ask verified humans

    Example session with this skill installed

    Check if this AI response is accurate based on the policy:
    Policy: Returns accepted within 30 days.
    AI Response: You can return it within 60 days.

    • Read your context and instructions
    • Compiled the ask verified humans

    I have submitted the verification task to Plures to check the AI response against the return policy. The human verifiers have reached a unanimous consensus.

    Verification Result: AI Accuracy Check

    FieldValue
    Task IDtask_88291_cv
    QuestionIs the AI's answer correct under the policy?
    StatusCompleted
    ConsensusUnanimous
    Selected OptionNo
    Vote Count3/3

    Result Summary
    The human verifiers unanimously agreed that the AI response is incorrect. The policy states a 30-day limit, while the AI incorrectly claimed a 60-day window.

    Next steps

    • Update the system prompt to emphasize the 30-day return window.
    • Correct the draft response before sending it to the customer.
    • Log this error in the evaluation dataset for future model fine-tuning.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    AI agents are confidently wrong more often than anyone notices. This skill gives yours a human in the loop.

    When your agent hits a judgment call where a mistake is costly, it sends the question to ID verified people through Plures and gets their consensus back, usually within a minute. Unanimous answers are safe to act on. Split answers go to you.

    Great for

    • Evals: is this AI answer correct, safe and on policy?
    • Training feedback: which model response is better?
    • Tone and intent in customer messages
    • Moderation edge cases
    • Matching: does a document, listing or screenshot match what it claims?

    What the skill does

    • Installs the free plures-mcp server
    • Tells your agent when to ask people, and when a lookup or test is enough
    • Shows it how to write a clear question with options and context
    • Acts on the result: unanimous answers go ahead, split answers come back to you
    • Keeps secrets and unneeded personal data out of every task

    Works with Claude Code, Cursor, Claude Desktop, VS Code and any MCP client.

    Pricing

    The skill is free. Try it with a free test key from plures.si, which returns simulated answers instantly. Live answers from real verified people are billed per decision.

    Needs a Plures API key. Test keys return simulated answers; live keys are billed per answer. Tasks are shown to verified strangers, so never send secrets or more personal data than the decision needs. Not for money or account access decisions, which belong to your own team. If people haven't answered within 240 seconds, the agent gets a task ID and checks back later.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 13 days ago

    • Free with an Agensi account

    Requires

    Plures MCP
    Node.js

    Permissions

    Network Access
    Listed13 days ago

    What's inside

    Frequently Asked Questions

    ID verified people on the Plures network. Each task goes to 1, 3, 5, 7 or 9 of them (3 by default), and your agent gets back the answer, how many agreed, and whether it is safe to act on.

    Anything that supports MCP: Claude Code, Cursor, Claude Desktop, VS Code, and agent frameworks like the OpenAI Agents SDK, LangChain and the Vercel AI SDK.

    Yes. Pass an image_url along with the question and any text context, and people judge the image directly.

    You will receive the core logic to integrate human-in-the-loop workflows, though you will need an active connection to the Plures MCP server to facilitate the actual human verification requests.

    If a consensus isn't reached immediately, the skill uses a task ID system to poll for results, allowing your agent to either wait or move on to other tasks and check back later.

    The skill includes response auditing tools that allow your agent to see if a vote was unanimous or split, enabling you to program specific 'safety-first' logic for disputed results.

    New in LLM Ops