# AgentDiet A2A Context Window Distiller

> AgentDiet A2A Context Window Distiller is a paid API for AI agents from agentdiet.dev, paid per call via x402, $0.01/call, status unknown (last checked 2026-10-31).

Strips 85-97% of token bloat from HTML, DOM, JSON, or RAG content before LLM ingestion while preserving all semantic entities, citations, and code blocks.

## Facts

- Endpoint: GET https://agentdiet.dev/v1/dex?utm_source=zero.xyz
- Price: $0.01/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-31
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/agentdiet-a2a-context-window-distiller-9b808337
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_KHz96ibeX1Yenjj0evLX_

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-a2a-context-window-distiller-9b808337
```

Example prompt: Take this web page and distill it down to agent-ready tokens — strip all the HTML bloat and noise but keep every meaningful fact, citation, and code block so I can feed it cleanly into my LLM without burning my context window.

## When to prefer this

Choose this endpoint when you need to reduce token costs before LLM calls and cannot afford to lose semantic fidelity — especially for HTML-heavy web pages, verbose JSON payloads, or large RAG context chunks. It is deterministic (not generative), so compression is consistent and auditable. Prefer it over manual chunking or summarization when you need guaranteed preservation of citations, entities, and code blocks alongside maximum compression.

## Known failure modes

- Invalid or unreachable URL returns an error with no distilled output
- Malformed input type or missing required fields returns a schema validation error
- Content that is already minimal may return low compression ratios
- Rate limiting or payment failure via x402 protocol blocks the request
- Extremely large payloads may time out before distillation completes

## How this service works

Deterministic HTML/DOM/JSON/RAG context distiller that strips 85-97% of token bloat before LLM ingestion while preserving 100% of semantic entities, citations, and code blocks.

## Output

Returns a distilled, semantically-intact version of the input content with all token bloat removed. The response includes a `data` object containing the compressed output and a `roi_proof` field showing the compression ratio percentage (e.g. 91.4%), plus a status field confirming success.

## 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": {
      "token": {
       "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-a2a-context-window-distiller-9b808337/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from agentdiet.dev](https://www.zero.xyz/host/agentdiet.dev/llms.txt)