CogDx Pre-Trade Audit
Cognitive verification layer for prediction market trades. Catches reasoning flaws before they become losses.
This is a template. The default signal is your agent's trade thesis —
the skill audits the reasoning quality before execution.
Remix it with stricter thresholds, additional bias checks, or custom fallacy detection.
The skill handles all the plumbing (API calls, trade execution, safeguards).
Your agent provides the reasoning to verify.
What it does
- 1. Takes your trade reasoning (thesis, confidence, market context)
- Runs cognitive diagnostics via CogDx API
- Returns verdict: PROCEED / REVIEW / REJECT
- Optionally executes trade if reasoning passes
Detects
- - Logical fallacies: anchoring, confirmation bias, sunk cost
- Calibration issues: overconfidence, underconfidence
- Reasoning gaps: missing evidence, circular logic
Usage
CODEBLOCK0
Environment Variables
- -
SIMMER_API_KEY - Required. Your Simmer API key. - INLINECODE1 - Optional. Wallet address for CogDx credits.
Thresholds
| Parameter | Default | Description |
|---|
| INLINECODE2 | 0.7 | Minimum reasoning quality score (0-1) |
| INLINECODE3 |
True | Block trade if CogDx API unavailable |
Why use this
Most trading losses come from bad reasoning, not bad data. This skill catches:
- - Trades based on anchoring (first number you saw)
- Confirmation bias (only seeing supporting evidence)
- Overconfidence (betting big on weak signals)
External verification you can't do yourself.
Credits
Built by Cerebratech — cognitive diagnostics for agents.
CogDx 交易前审计
预测市场交易的认知验证层。在推理缺陷变成损失之前将其捕获。
这是一个模板。 默认信号是你的代理的交易论点——
该技能在执行前审计推理质量。
使用更严格的阈值、额外的偏见检查或自定义谬误检测进行重新混合。
该技能处理所有底层工作(API调用、交易执行、安全防护)。
你的代理提供待验证的推理。
功能
- 1. 获取你的交易推理(论点、置信度、市场背景)
- 通过CogDx API运行认知诊断
- 返回裁决:继续 / 审查 / 拒绝
- 如果推理通过,可选择执行交易
检测内容
- - 逻辑谬误:锚定效应、确认偏误、沉没成本
- 校准问题:过度自信、信心不足
- 推理漏洞:证据缺失、循环论证
使用方法
python
from cogdxpretradeaudit import auditand_trade
result = auditandtrade(
market_id=0x1234...,
side=yes,
amount=10.0,
reasoning=基于SEC会议纪要,BTC ETF获批可能性较大...,
confidence=0.85,
min_validity=0.7, # 继续交易的最低推理质量
live=False # 默认进行试运行
)
if result[approved]:
print(f交易已执行: {result[trade_id]})
else:
print(f交易被阻止: {result[issues]})
环境变量
- - SIMMERAPIKEY - 必填。你的Simmer API密钥。
- COGDX_WALLET - 可选。CogDx积分的钱包地址。
阈值
| 参数 | 默认值 | 描述 |
|---|
| minvalidity | 0.7 | 最低推理质量评分(0-1) |
| blockon_error |
True | 若CogDx API不可用则阻止交易 |
为何使用此功能
大多数交易损失源于糟糕的推理,而非糟糕的数据。该技能可捕获:
- - 基于锚定效应(你看到的第一个数字)的交易
- 确认偏误(只看到支持性证据)
- 过度自信(对弱信号下重注)
你无法自行完成的外部验证。
致谢
由 Cerebratech 构建——为代理提供认知诊断。