# AgentDiet RAG Poison Shield

> AgentDiet RAG Poison Shield is a paid API for AI agents from status.agentdiet.dev, paid per call via x402, $0.005/call, status unknown (last checked 2026-10-22).

Scans text input to detect and flag RAG poisoning attempts that could corrupt AI agent context windows

## Facts

- Endpoint: GET https://status.agentdiet.dev/v1/rag-poison-shield?utm_source=zero.xyz
- Price: $0.005/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-22
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/agentdiet-rag-poison-shield-383096e7
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_0H5Ka3-O3C-kRT61TN1YD

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability agentdiet-rag-poison-shield-383096e7
```

Example prompt: Before I pass this retrieved document chunk into my agent's context, run it through the RAG poison shield to check if it contains any adversarial injection or manipulation attempts.

## When to prefer this

Choose this endpoint when your AI agent retrieves content from untrusted or semi-trusted external sources (web scraping, user-submitted documents, third-party APIs) and needs to validate that content before injecting it into a context window. It is particularly valuable in agentic RAG pipelines where malicious actors could embed adversarial instructions inside retrieved documents. Prefer this over generic content moderation when the threat model is specifically prompt injection or RAG manipulation rather than general toxicity.

## Known failure modes

- Empty or missing 'text' query parameter returns an error or empty analysis
- Non-text binary input may produce unreliable shield results
- Payment failure via x402/USDC may return 402 status blocking access
- Very long text inputs exceeding context limits may be truncated or rejected
- False negatives: sophisticated poisoning attempts may not be detected

## How this service works

Deterministic B2A/A2A/MCP/x402/MPP/L402 execution and intelligence node at https://status.agentdiet.dev settling in USDC on Base.

## Output

Returns a JSON object with a status field ('ok' or error), a data payload containing analysis results, and an roi_proof object showing the compression_ratio_pct (e.g. 91.4%) achieved — indicating how much of the input was flagged or distilled as potentially adversarial or redundant.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "$schema": "https://json-schema.org/draft/2020-12/schema",
 "required": [
  "input"
 ],
 "properties": {
  "input": {
   "type": "object",
   "required": [
    "type",
    "method"
   ],
   "properties": {
    "type": {
     "type": "string",
     "const": "http"
    },
    "method": {
     "enum": [
      "GET",
      "POST"
     ],
     "type": "string"
    },
    "queryParams": {
     "type": "object",
     "properties": {
      "text": {
       "type": "string"
      }
     }
    }
   },
   "additionalProperties": false
  },
  "output": {
   "type": "object",
   "required": [
    "type"
   ],
   "properties": {
    "type": {
     "type": "string"
    },
    "example": {
     "type": "object"
    }
   }
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "json",
 "example": {
  "data": {},
  "status": "ok",
  "roi_proof": {
   "compression_ratio_pct": 91.4
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/agentdiet-rag-poison-shield-383096e7/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from status.agentdiet.dev](https://www.zero.xyz/host/status.agentdiet.dev/llms.txt)