Game Theory Strategist

• Outcome: Review
• Latest audit: Sep 16, 2026 · 4:51 AM UTC
• Version: 1.0.1

Overview

The skill has no hidden code access, but its broad activation and under-guarded negotiation guidance can push manipulative or escalatory tactics in sensitive personal and workplace situations.

Review before installing if you expect this skill to be used around family, workplace, legal, financial, or relationship disputes. It should be constrained to transparent, voluntary, non-coercive strategic analysis, and users should avoid treating its outputs as professional advice or as permission to pressure counterparties. Also note that the required output headings are hardcoded in Spanish.

A.I.G

Vulnerability Patterns

• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally
• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (13)

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 93% confidence

Finding

The description promises a comprehensive game theory analyzer for many strategic scenarios, including equilibrium concepts and visual outputs. The code instead performs a specific mechanism-design check: for each agent, it compares truthful participation against outside options and deceptive behavior to determine IR and IC, then suggests adjustments. While this is loosely related to mechanism design and conflict/negotiation contexts, it is materially narrower than the declared functionality and lacks the advertised core capabilities and outputs. There are no concerning undeclared permissions or resource accesses; the mismatch is primarily purpose and capability scope.

Content

No source excerpt is available for this finding.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 92% confidence

Finding

The declared description is broad multi-agent game-theoretic analysis centered on interactions where outcomes depend on multiple decision-makers. The supplied code instead performs regret matching on exogenously supplied utility vectors for a single decision-maker across repeated simulations. While CFR is related to game theory, this implementation does not represent opponents, reaction functions, game trees, payoff matrices, or equilibrium concepts named in the description. Its primary purpose is closer to stochastic decision weighting under regret than full strategic interaction analysis. The absence of visual generation is also a notable gap relative to the declared outputs.

Content

No source excerpt is available for this finding.

Vague Triggers

High

• Category: Not specified by scanner
• Confidence: 97% confidence

Finding

The manifest uses very broad activation triggers like negotiation, conflict, career pivots, and common questions such as 'what should I do,' which can cause the skill to activate in many unrelated contexts. Over-broad triggering is dangerous because it can hijack normal conversations, inject irrelevant strategic framing into sensitive situations, and increase the chance of the wrong tool being used for legal, mental-health, relationship, or other high-stakes advice.

Content

No source excerpt is available for this finding.

Ae1

High

• Category: analysis-evasion
• Confidence: 100% confidence

Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 97)May include surrounding context.

md

- `evals/evals.json` — 5 benchmark scenarios with expected outputs. Use to verify skill quality.

Natural-Language Policy Violations

High

• Category: Not specified by scanner
• Confidence: 98% confidence

Finding

This section endorses brinkmanship, shared-risk escalation, and irreversible commitment devices as effective negotiation tools, including examples of raising the probability of mutual harm and 'burning bridges' to make threats credible. In the context of a strategy skill for personal, workplace, and conflict situations, this materially increases the risk of coercion, retaliation, unsafe escalation, and harmful real-world actions.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 94% confidence

Finding

The output format mandates headings like "CLASIFICACIÓN DEL JUEGO" and "EQUILIBRIO DE NASH," which imposes a specific language in the generated artifact. There is no opt-in, locale selection, or documented justification for requiring Spanish output.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 94% confidence

Finding

The expected_sections field requires outputs using Spanish headers such as "CLASIFICACIÓN DEL JUEGO" and "VEREDICTO ESTRATÉGICO". This is a natural-language locale constraint in the skill content, but the file does not indicate user opt-in, multilingual support, or a documented reason for enforcing Spanish.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 93% confidence

Finding

This scenario's expected_sections again mandates Spanish section titles, reinforcing a Spanish-only output policy. Because the file is JSON and applies across scenarios, this looks like a built-in locale requirement rather than a user-selected preference.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 92% confidence

Finding

The expected_sections list continues to prescribe Spanish headings while the rest of the scenario text is in English. That mismatch suggests the skill enforces a specific output language without user opt-in or clear justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 92% confidence

Finding

The scenario requires Spanish section names in expected_sections, which constitutes a language policy embedded in the file. No surrounding metadata explains that the skill is region-specific or asks the user to choose a preferred language.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 92% confidence

Finding

The expected_sections field again hardcodes Spanish labels, indicating a consistent locale restriction across the evaluation file. Under the policy, forcing a language without opt-in is a reportable natural-language policy issue.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 95% confidence

Finding

The text explicitly advises exploiting cognitive biases in negotiation by framing offers around loss aversion and using high anchors when the speaker has an information advantage. In a coaching skill meant for real-world negotiation and conflict scenarios, this goes beyond neutral description into prescriptive manipulation without consent, safeguards, or ethical constraints, which can enable coercive or deceptive behavior.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium

• Category: Not specified by scanner
• Confidence: 94% confidence

Finding

The suggest_fix method recommends interventions such as 'lower the outside option via timing/urgency' and adding monitoring or penalties without any safety boundaries, consent framing, or ethical caveats. In a skill explicitly meant for negotiation, conflicts, co-founder disputes, and household coordination, these suggestions can normalize manipulative pressure tactics and coercive relationship dynamics rather than neutral analysis.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.