Skip to content
Back to skills

Validate Predictions

ASecurity

Review and validate pending predictions against current market data. Usage: /validate-predictions

  • 32 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added May 26, 2026
ai-agents

Works with

  • mcp

Security analysis

A100/100

Scanned May 27, 2026

npx -y skills add hugoguerrap/crypto-claude-desk --skill validate-predictions --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Validate Predictions?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Validate Predictions
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/hugoguerrap-validate-predictions/badge)](https://www.skillsdirectory.com/skills/hugoguerrap-validate-predictions)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: validate-predictions
description: Review and validate pending predictions against current market data. Usage: /validate-predictions
user-invocable: true
---

# Validate Predictions

Review all pending predictions and check them against current market data.

## Workflow

### Step 0: Auto-Find Expired Predictions
Delegate using the Task tool with `subagent_type: general-purpose` and `model: opus`:

"You are the learning-agent. Read agents/learning-agent.md for your analysis framework.
First, use get_crypto_prices() from crypto-data MCP to get current prices for major coins (bitcoin, ethereum, solana, etc.).
Then call find_expired_predictions(current_prices='{\"BTC/USDT\": ..., \"ETH/USDT\": ...}') from crypto-learning-db to discover predictions whose timeframe has passed.
For each expired prediction, reason about how close it was and validate with an NL evaluation using validate_prediction().
Do NOT use the Edit tool."

### Step 1: Check Remaining Pending Predictions
Delegate using the Task tool with `subagent_type: general-purpose` and `model: opus`:

"You are the learning-agent. Call query_predictions(status='pending') from crypto-learning-db for predictions still within their timeframe.
For each prediction:
1. Use get_exchange_prices(symbol=...) from crypto-exchange MCP to check current price
2. Compare current price against the prediction's target_value
3. Report current progress toward or away from target
Do NOT use the Edit tool."

### Step 2: Present Results
Show a summary table:

```
## Prediction Validation Report

### Resolved This Check
| ID | Agent | Prediction | Target | Actual | Result |
|----|-------|-----------|--------|--------|--------|

### Still Pending
| ID | Agent | Prediction | Target | Current | Progress | Expires |
|----|-------|-----------|--------|---------|----------|---------|

### Overall Accuracy
- Total predictions: X
- Correct: X (X%)
- Incorrect: X (X%)
- Pending: X

### Track Record by Setup Type
| Setup Type | Total | Correct | Accuracy | Trend |
|-----------|-------|---------|----------|-------|
```

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.