Tool detail

Pocket Network

Classifies agent-bound text as safe, suspicious or malicious for prompt injection and returns the verdict with matched patterns.

Categories: Agent trust safety

Use when

screen incoming text for prompt injection before an agent reads it
classify a webpage or tool response as safe before acting on it
check user-supplied text for role-override or injection patterns

Not for

not a guarantee of safety - a ruleset-based classifier, not a formal proof
sends the inspected text to Pocket Network for classification

Inputs

POST JSON body: { text: string } (required)

Admitted input schema

{
  "type": "object",
  "properties": {
    "text": {
      "type": "string",
      "minLength": 0,
      "maxLength": 4096,
      "description": "The agent-bound text to inspect."
    }
  },
  "required": [
    "text"
  ],
  "additionalProperties": false,
  "description": "Text to classify for prompt injection."
}

Endpoint

Method: POST
URL: https://agent.pocket.network/v1/prompt-injection-detect

Payment

Protocol: x402
Listed price: 0.005 USDC per call on Base
Payment method: exact
Network: eip155:8453
Asset: 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913

Listed terms are catalog metadata. The caller supplies inputs and calls the tool outside Onelayer. If payment is required, the caller uses a compatible wallet to pay.

Source

Directory: Coinbase x402 Bazaar
Record: POST https://agent.pocket.network/v1/prompt-injection-detect
Retrieved: 2026-09-30T20:37:07.382Z
Catalog file SHA-256: 6837ccddf886f0007cf3727ea9d27faaa5047b51b4f425345c554ce1c271a524
Catalog built: 2026-09-30T22:31:58.034Z