Developer platform

Build on the trust layer.

Give any agent a verified-human identity and a provable conscience. Mint a key, call the governed API, connect the MCP server — every action returns a hash-chained, Ed25519-sealed, on-chain-anchored receipt anyone can verify.

—

Missions

—

Completed

MCP tools

480

World Chain

API keys

Keys are scoped to one verified human. Prove you’re human once with World ID, then mint as many keys as you need — each is shown a single time and stored only as a hash.

Preview mode — verify to demo the full flow in any browser.

Quickstart

Three calls: create a mission, execute it to a sealed receipt, and verify the receipt — including its World Chain anchor.

# 1 · Create a governed mission
MID=$(curl -s https://agent.aweblabs.ai/api/v1/missions \
  -H "Authorization: Bearer $AWEB_AGENT_KEY" \
  -H "content-type: application/json" \
  -d '{"goal":"Research 2026 stablecoin trends, summarize top 3"}' | jq -r .missionId)

# 2 · Execute → sealed + on-chain-anchored receipt
curl -s -X POST https://agent.aweblabs.ai/api/v1/missions/$MID/execute \
  -H "Authorization: Bearer $AWEB_AGENT_KEY" -d '{}'

# 3 · Verify the receipt (public, no key) — checks the anchor on World Chain
curl -s "https://agent.aweblabs.ai/api/v1/receipts/$MID/verify?onchain=1"

REST API

Base URL https://agent.aweblabs.ai. Authenticate writes with Authorization: Bearer sk-aweb-…. Receipt reads are public and CORS-open.

POST

/api/v1/missions

Create + plan a governed mission

POST

/api/v1/missions/:id/execute

Execute → sealed + anchored receipt (World ID proof for sensitive steps)

GET

/api/v1/missions/:id

Mission status + plan

GET

/api/v1/receipts/:id

Full verifiable receipt + attestation · public

GET

/api/v1/receipts/:id/verify?onchain=1

Attestation only — reads the anchor calldata on-chain · public

GET

/api/v1/openapi.json

OpenAPI 3.1 specification

# 1 · Create a governed mission
MID=$(curl -s https://agent.aweblabs.ai/api/v1/missions \
  -H "Authorization: Bearer $AWEB_AGENT_KEY" \
  -H "content-type: application/json" \
  -d '{"goal":"Research 2026 stablecoin trends, summarize top 3"}' | jq -r .missionId)

# 2 · Execute → sealed + on-chain-anchored receipt
curl -s -X POST https://agent.aweblabs.ai/api/v1/missions/$MID/execute \
  -H "Authorization: Bearer $AWEB_AGENT_KEY" -d '{}'

# 3 · Verify the receipt (public, no key) — checks the anchor on World Chain
curl -s "https://agent.aweblabs.ai/api/v1/receipts/$MID/verify?onchain=1"

MCP server

Connect any MCP client (Claude, your own agent) over Streamable HTTP. Drop this into your MCP config and swap in a key — the same governed surface, callable by machines.

{
  "mcpServers": {
    "aweb-world-agent": {
      "type": "streamable-http",
      "url": "https://agent.aweblabs.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer sk-aweb-...",
        "MCP-Protocol-Version": "2025-06-18"
      }
    }
  }
}

Tools (6) · protocol 2025-06-18

world_mission_createworld_mission_executeworld_mission_statusworld_receipt_getworld_receipt_verifyworld_stats

Verify receipts

Every receipt is independently checkable — integrity (hash chain), authenticity (Ed25519 seal), and anchor (the sealed root committed on World Chain). With ?onchain=1 the API reads the tx calldata back and confirms it equals the root. Zero trust in us required.

Integrityhash chain

AuthenticityEd25519

AnchorWorld Chain

OpenAPI · Open source · World proves a human is behind the agent · Aweb proves it behaved