# Agent Council (Deep Mode)

> Agent Council (Deep Mode) is a paid API for AI agents from council.cyberwarex.com, paid per call via x402, $0.03/call, status unknown (last checked 2026-10-12).

Submits a decision question to 4 AI models in two debate rounds, returning a synthesized verdict with individual positions, confidence scores, dissent, and what would flip the conclusion.

## Facts

- Endpoint: GET https://council.cyberwarex.com/council-deep?utm_source=zero.xyz
- Price: $0.03/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-12
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/agent-council-deep-mode-5271853a
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_DD3gHsYrBjCETsCh77JQe

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 agent-council-deep-mode-5271853a
```

Example prompt: Put this question to your deep 4-model council with two rounds of debate: should we launch the new pricing tier at $49/month targeting SMBs, or hold until we have 10 more enterprise logos — we have 6 months of runway and 200 beta users asking for it?

## When to prefer this

Choose this endpoint over the quick (3-model, 1-round) council when the decision is high-stakes and being wrong is expensive — it provides two rounds of cross-examination, so seats revise after seeing each other's positions, producing a verdict that has survived internal challenge. Prefer this over the grounded council when you already have the facts and need deliberation, not research. Use when you want to see dissent that held, not just the winning argument.

## Known failure modes

- Missing required 'q' query parameter returns an error
- Question exceeding 2000 characters is rejected
- Payment failure or missing x402 payment header returns 402
- Upstream AI model timeout may increase elapsed_sec or cause partial panel failure
- Vague or non-decision questions may produce low-confidence verdicts with wide dissent
- Rate limiting or service unavailability returns 5xx

## How this service works

The deep council: 4 different AI models, two rounds. Each seat answers, then sees the other seats' positions and revises - so you get a verdict that survived cross-examination, with changed_after_debate on each seat and the dissent that held. For decisions where being wrong is expensive.

## Output

Returns a structured object with: a synthesized 'verdict' string from the chair, an array of 'panel' entries (each with the model, its position, confidence, and whether it changed_after_debate), a 'dissent' array of real disagreements that survived two rounds, a 'consensus' array of points all seats agreed on, a 'watch_for' list of conditions that would flip the verdict, an integer 'confidence' (0-100, lower when seats disagreed), 'rounds' (2), 'seats' (4), 'mode' ('deep'), and 'elapsed_sec'.

## 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",
      "HEAD",
      "DELETE"
     ],
     "type": "string"
    },
    "queryParams": {
     "type": "object",
     "required": [
      "q"
     ],
     "properties": {
      "q": {
       "type": "string",
       "maxLength": 2000,
       "description": "The question to put to the council (required, max 2000 chars). A decision question works best: include the options and the constraints."
      }
     },
     "additionalProperties": false
    }
   },
   "additionalProperties": false
  },
  "output": {
   "type": "object",
   "required": [
    "type"
   ],
   "properties": {
    "type": {
     "type": "string"
    },
    "example": {
     "type": "object",
     "properties": {
      "mode": {
       "enum": [
        "quick",
        "deep"
       ],
       "type": "string"
      },
      "panel": {
       "type": "array",
       "items": {
        "type": "object"
       },
       "description": "each seat's position, confidence and the model that argued it."
      },
      "seats": {
       "type": "integer"
      },
      "rounds": {
       "type": "integer"
      },
      "dissent": {
       "type": "array",
       "items": {
        "type": "object"
       },
       "description": "real disagreements, as {seat, point} - empty when the panel agreed."
      },
      "verdict": {
       "type": "string",
       "description": "the council's answer, synthesized by the chair."
      },
      "question": {
       "type": "string"
      },
      "consensus": {
       "type": "array",
       "items": {
        "type": "string"
       }
      },
      "watch_for": {
       "type": "array",
       "items": {
        "type": "string"
       },
       "description": "what would flip the verdict."
      },
      "confidence": {
       "type": "integer",
       "description": "0-100, lower when the seats disagreed."
      },
      "elapsed_sec": {
       "type": "number"
      }
     }
    }
   }
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/agent-council-deep-mode-5271853a/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from council.cyberwarex.com](https://www.zero.xyz/host/council.cyberwarex.com/llms.txt)