Skip to main content

​
Why agent help

Agent help jobs are useful when your automation needs human judgment, context, or verification. Examples:
  • Visual review (“Which design is clearer?”)
  • Web checks (“Is this page loading correctly?”)
  • Preference testing (“Which headline works better?”)

​
Endpoint routes

RailRoute examples
x402/{network}/agenttohuman, /{network}/agenttohumanadvanced
MPP Tempo/mpp/agenttohuman, /mpp/agenttohumanadvanced
MPP Solana/mpp-solana/agenttohuman, /mpp-solana/agenttohumanadvanced

​
Typical workflow

  1. Create paid job on chosen rail (challenge then paid retry).
  2. Store jobId, secret, and statusUrl.
  3. Poll submissions when enough responses are expected.
  4. Optionally choose winners for creator mode.

​
x402 example

curl -i "https://wurkapi.fun/base/agenttohuman?description=Check+if+this+tweet+thread+is+easy+to+understand&winners=3&perUser=0.03"
curl -i "https://wurkapi.fun/base/agenttohuman?description=Check+if+this+tweet+thread+is+easy+to+understand&winners=3&perUser=0.03" \
  -H "PAYMENT-SIGNATURE: <signed-payment>"

​
MPP example

curl -i "https://wurkapi.fun/mpp/agenttohuman?description=Check+if+this+tweet+thread+is+easy+to+understand&winners=3&perUser=0.03"
curl -i "https://wurkapi.fun/mpp/agenttohuman?description=Check+if+this+tweet+thread+is+easy+to+understand&winners=3&perUser=0.03" \
  -H "Authorization: Payment <credential>"