Security audit

Modern Chanakya

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed advice-and-reasoning skill with broad life-strategy framing, not a skill that installs backdoors, handles credentials, or performs hidden actions.

Install this only if you want a broad strategic-advice lens. For medical, legal, financial, relationship-safety, or crisis decisions, treat it as framing support and seek qualified help where appropriate.

A.I.G

Vulnerability Patterns

Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration

Findings (15)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 98% confidence
Finding: The declared purpose centers on substantive reasoning: interpreting classical Chanakya sources into modern advice and maintaining an indexed knowledge system. The actual code only constructs a ready-to-fill markdown benchmark block via command-line arguments. Its primary function is evaluation/scaffolding infrastructure, not content interpretation, strategic guidance, or knowledge retrieval. This is a material description-behavior mismatch because the implemented capability is a tooling script for benchmark creation, which is not merely a supporting detail of the declared advisory skill.
Content: No source excerpt is available for this finding.

Direct Prompt Extraction

High

Category: System Prompt Leakage
Confidence: 85% confidence
Finding: Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
Content: Scanner excerpt · references/frameworks/ethics-in-unstable-systems.md (reported line 45)May include surrounding context.
md
- Which skills gain value in unstable periods? - Which comforts are being mistaken for durable strength? ## Output rule When giving foresight advice: - distinguish likely vs possible - surface hidden risks

Direct Prompt Extraction

High

Category: System Prompt Leakage
Confidence: 85% confidence
Finding: Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
Content: Scanner excerpt · references/frameworks/foresight-and-risk.md (reported line 31)May include surrounding context.
md
- Which skills gain value in unstable periods? - Which comforts are being mistaken for durable strength? ## Output rule When giving foresight advice: - distinguish likely vs possible - surface hidden risks

Vague Triggers

High

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The trigger principle uses very broad concepts like strategic judgment, negotiation, foresight, and balancing risk, which appear in ordinary user requests across many domains. Such overbroad matching can cause the skill to activate unexpectedly, override more appropriate skills, and route sensitive or nuanced queries into a framework the user did not explicitly request.
Content: No source excerpt is available for this finding.

Vague Triggers

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The natural trigger clusters contain many common phrases such as burnout, apology message, compare flights, difficult boss, and best grocery option that overlap heavily with everyday speech. This creates a substantial risk of accidental triggering on unrelated conversations, broadens the effective capability surface far beyond the manifest, and may lead to inappropriate guidance in sensitive personal, financial, or relationship contexts.
Content: No source excerpt is available for this finding.

Anti-Refusal Statement

High

Category: Anti-Refusal
Confidence: 85% confidence
Finding: Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.
Content: Scanner excerpt · research/articles/kalzumeus-salary-negotiation-note.md (reported line 19)May include surrounding context.
md
- the skill of asking is economically high leverage ## Modern Chanakya translation - Do not moralize salary negotiation as greed. - Asking well is part of strategic adulthood. - The user should be trained to separate self-worth panic from practical negotiation behavior. - Better compensation is often won by calm framing, not by aggression.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The trigger guidance says to use the framework whenever the user asks about 'the future, uncertainty, instability, or hidden risks,' which covers a very wide range of ordinary conversations. It does not define boundaries, specific trigger phrases, or exclusion conditions, so the skill could be invoked in contexts where it is not intended.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium

Category: Not specified by scanner
Confidence: 89% confidence
Finding: The trigger library expands the skill from interpreting Chanakya-style principles into broad generic advice domains such as relationships, health, shopping, flights, and internet search. This can cause the skill to activate for requests outside the manifest’s declared scope, leading to unintended delegation, misleading authority, and reduced user transparency about what specialized capability is being invoked.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 91% confidence
Finding: The short-form invocation examples include phrases like 'Give me the strategic answer' and 'What is the smartest move?' that are too generic to safely distinguish this skill from normal assistant behavior. If reused as trigger patterns, they can silently widen activation and reduce user awareness that a specialized interpretive framework is being applied.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The prompt aliases are extremely short and generic, such as 'MC: analyze this' and 'MC: risk view', which increases the chance the skill is invoked unintentionally from ordinary user text or overlapping workflows. In an agent environment, accidental invocation can cause the wrong skill to steer responses, introduce unintended context switching, or apply strategic framing where it was not explicitly requested.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The invocation guidance is broad enough to capture a large range of ordinary personal and professional requests, which can cause over-invocation of the skill outside a narrowly justified scope. In an agent system, this increases the chance that a strategic persona or decision framework is applied where simpler, safer, or domain-specialized handling would be more appropriate, leading to misrouting and unintended influence over user decisions.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The large set of generic trigger examples spans career, health, wealth, relationships, negotiation, startups, social presence, and daily purchasing without meaningful constraints. This makes the skill likely to match many common-life queries and can normalize the skill as a default advisor for sensitive domains, increasing the risk of overreach, poor task selection, and advice being delivered with undue authority.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 89% confidence
Finding: The mode names are broad, natural-language phrases such as 'risk view mode', 'wealth mode', and 'emergency mode', which increases the chance that ordinary user text could accidentally match or partially invoke a mode. In a skill that steers advice and framing, unintended activation can alter the assistant's behavior, emphasis, or tone without the user explicitly intending it, creating prompt-routing ambiguity.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 81% confidence
Finding: Line L03 directs the assistant to answer in the "strongest Modern Chanakya voice," which imposes a particular linguistic/cultural style by default. Under the policy rule, forcing a specific language or locale style without user choice can be a natural-language policy violation.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

Low

Category: Not specified by scanner
Confidence: 80% confidence
Finding: The social media and content-strategy triggers push the skill into general growth-marketing and audience-building use cases that are only weakly tied to the stated Chanakya interpretation role. While not directly harmful on its own, this broadening increases the chance of incorrect or overconfident use of the skill in unrelated contexts.
Content: No source excerpt is available for this finding.