Oraclaw Decide

Decision intelligence for AI agents. Analyze options, map decision dependencies with PageRank, detect when information sources conflict, and find the choices...

Security Scan

The skill claims to call an external decision service (paid) but the runtime instructions do not explain how the declared ORACLAW_API_KEY is used or where network calls go, creating an incoherence you should resolve before trusting sensitive credentials.

Purpose & Capability

The skill description and pricing imply an external SaaS (paid per call) which reasonably could require ORACLAW_API_KEY. However, the SKILL.md provides no network endpoints, API call examples, or instructions that reference the API key. Requiring a key without documenting its use is inconsistent.

Instruction Scope

SKILL.md is an instruction-only spec describing capabilities and input formats but contains no explicit runtime steps for calling an external API, nor any guidance on using ORACLAW_API_KEY. That omission grants the agent broad discretion (e.g., to contact unknown endpoints) without transparency about what data would be sent.

Install Mechanism

No install spec or code files are present (instruction-only). This minimizes disk-write risks — there is nothing being downloaded or installed by the skill itself.

Credentials

Only one credential (ORACLAW_API_KEY) is requested, which is proportionate if the skill calls an external Oraclaw API. But the SKILL.md never references that variable or shows authentication flows. The pricing/payment note further implies external billing. Requesting a secret without documenting its scope, endpoints, or what data is transmitted is a red flag.

Persistence & Privilege

always is false and the skill is not granted forced persistent presence. It does allow normal autonomous invocation (platform default), which is expected — no extra privileges are requested.

What to consider before installing

This skill appears to be a front-end for a paid decision-making API, but the runtime instructions don't say how the API key is used or where requests are sent. Before installing or providing ORACLAW_API_KEY you should: 1) Ask the publisher for API endpoint URLs, example requests/responses, and a privacy policy showing what user data is transmitted and stored. 2) Verify the homepage and publisher reputation and confirm payment details (the SKILL.md mentions $0.05/analysis). 3) If you test it, use a least-privilege or expendable API key and only run non-sensitive data. 4) Prefer to see explicit SKILL.md examples that show where the key is read and how authentication happens; absence of that detail is the core incoherence here. If the maintainer cannot provide clear documentation, treat the skill as risky to give credentials to.

OraClaw Decide — Decision Intelligence for Agents

You are a strategic decision agent that uses graph analysis, convergence scoring, and optimization to make and analyze decisions.

When to Use This Skill

Use this when the user or another agent needs to:

• Choose the best option from competing alternatives
• Map dependencies between decisions and find bottlenecks
• Check if multiple information sources agree or conflict
• Identify which decisions have the highest ripple effect
• Find the critical path through a complex project

Tools Available

optimize_bandit — Choose the Best Option

Given options with historical performance, select the one with highest expected value.

optimize_contextual — Context-Aware Decisions

Choose differently based on the current situation (time pressure, stakes, complexity).

analyze_decision_graph — Map & Analyze Decision Networks

Feed in decisions as nodes and relationships as edges. Get back:

• PageRank: Which decisions are most influential?
• Communities: Which decisions cluster together?
• Bottlenecks: What's blocking everything?
• Critical path: Shortest route from start to goal

score_convergence — Are Your Sources Agreeing?

When you have multiple forecasts, estimates, or opinions — score how much they agree. Detects outliers automatically.

Example: Project Decision Analysis

{
  "nodes": [
    { "id": "hire", "type": "decision", "label": "Hire senior dev", "urgency": "critical", "confidence": 0.4, "impact": 0.9, "timestamp": 1711350000 },
    { "id": "ship", "type": "goal", "label": "Ship v2.0", "urgency": "critical", "confidence": 0.5, "impact": 1.0, "timestamp": 1711350000 },
    { "id": "fundraise", "type": "decision", "label": "Start fundraise", "urgency": "high", "confidence": 0.6, "impact": 0.8, "timestamp": 1711350000 }
  ],
  "edges": [
    { "source": "hire", "target": "ship", "type": "enables", "weight": 0.9 },
    { "source": "ship", "target": "fundraise", "type": "enables", "weight": 0.8 }
  ],
  "sourceGoal": "hire",
  "targetGoal": "fundraise"
}

Rules

1. For graph analysis: nodes need all required fields (id, type, label, urgency, confidence, impact, timestamp)
2. Edge weights should be 0-1 (higher = stronger relationship)
3. Convergence scoring works best with 3+ sources
4. When sources disagree significantly (spread > 2000 bps), investigate the outlier before deciding

Pricing

$0.05 per analysis call (USDC on Base via x402). Free tier: 100 decisions/month with API key.