# Augur

> Deterministic Base contract admission control for agents on Base. Returns an effective decision, policy recommendation, supporting findings, and a 0-100 score. Pay $0.10/call via x402 in USDC on Base.

## What It Does

Augur fetches on-chain bytecode for a Base mainnet smart contract (EIP-155:8453) and runs 8 deterministic detectors to produce an effective decision, policy recommendation, supporting findings, and a composite score from 0 (safe) to 100 (critical).
Use Augur before your agent buys, routes funds, approves, pays, or interacts. Call before pay. Call before approve. Call before interact. Augur is deterministic preflight for Base contract actions.
A `safe` result means no major bytecode-level risk signals were detected in this scan, not that the contract is audited or guaranteed safe.

## How to Call

```
GET https://augurrisk.com/analyze?address=0x4200000000000000000000000000000000000006
```

Payment: Include a `PAYMENT-SIGNATURE` header with an x402 payment proof ($0.10 USDC on Base). Any x402-compatible HTTP client handles this automatically.

## First Successful Paid Call

Use this exact Base WETH request as the canonical integration check:

```http
GET https://augurrisk.com/analyze?address=0x4200000000000000000000000000000000000006
PAYMENT-SIGNATURE: <x402-payment-proof>
```

Expected first success: HTTP `200` JSON with `decision: "allow"`, `level: "safe"`, and `score: 0`.
If the request is missing `address`, appends another path after the address, or uses a malformed address, Augur returns `422` before payment.

## Action-Aware Example: Approve

Augur also accepts a narrow action-aware context for `approve` on Base:

```
GET https://augurrisk.com/analyze?address=0x4200000000000000000000000000000000000006&action=approve&spender=0x1111111111111111111111111111111111111111&chain=base
```

The top-level `decision` is the effective gate to branch on. For action-aware requests, `contract_decision` preserves the contract-only result while `action_context` plus `action_evaluation` show the action-specific details.
That means a clean contract can return `contract_decision: "allow"` while the primary `decision` for an approval returns `warn`.

## Example Response

```json
{
  "address": "0x4200000000000000000000000000000000000006",
  "score": 0,
  "level": "safe",
  "decision": "allow",
  "contract_decision": "allow",
  "recommended_policy": {
    "action": "allow",
    "summary": "Allow by default for first-pass automation. Continue only if this matches your broader strategy and trust model.",
    "reason_codes": []
  },
  "bytecode_size": 4632,
  "findings": [],
  "category_scores": {}
}
```

## Action-Aware Example Response

```json
{
  "address": "0x4200000000000000000000000000000000000006",
  "score": 0,
  "level": "safe",
  "decision": "warn",
  "contract_decision": "allow",
  "recommended_policy": {
    "action": "warn",
    "summary": "Allow with caution only if this workflow explicitly expects the approval. Keep the spender on an allowlist and the approval scope narrow.",
    "reason_codes": [
      "action_approve_requested"
    ]
  },
  "bytecode_size": 4632,
  "findings": [],
  "category_scores": {},
  "action_context": {
    "action": "approve",
    "spender": "0x1111111111111111111111111111111111111111",
    "chain": "base"
  },
  "action_evaluation": {
    "decision": "warn",
    "recommended_policy": {
      "action": "warn",
      "summary": "Allow with caution only if this workflow explicitly expects the approval. Keep the spender on an allowlist and the approval scope narrow.",
      "reason_codes": [
        "action_approve_requested"
      ]
    }
  }
}
```

Current limit: action-aware V1 only supports `action=approve` on Base. If no spender allowlist is configured, spender trust is treated as `unchecked`.

## Risk Levels

- **safe** (0-15): No major bytecode-level risk signals detected in this scan; measurement only, not a guarantee. Branch on `decision`, not `level`
- **low** (16-35): Limited bytecode-level concerns detected; review context
- **medium** (36-55): Notable risks, review before interacting
- **high** (56-75): Significant risks detected
- **critical** (76-100): Severe risks, avoid interaction

## Links

- [Skill Doc](https://augurrisk.com/skill.md)
- [MCP Setup](https://augurrisk.com/mcp)
- [MCP Package](https://www.npmjs.com/package/augurrisk-mcp)
- [OpenAPI Spec](https://augurrisk.com/openapi.json)
- [A2A Agent Card](https://augurrisk.com/.well-known/agent-card.json)
- [AI Plugin Manifest](https://augurrisk.com/.well-known/ai-plugin.json)
- [x402 Discovery](https://augurrisk.com/.well-known/x402)
- [API Catalog](https://augurrisk.com/.well-known/api-catalog)
- [Full Documentation](https://augurrisk.com/llms-full.txt)