ClawHub
Back to skill

Security audit

Fictional Companion Forge

Security checks for vulnerabilities and agentic risk

Overview

This is a markdown-only character companion generator with no hidden execution, credential access, network payloads, or persistence beyond generated companion files.

Install only if you want a skill that helps generate fictional companion-agent profiles. Be aware that its trigger wording is broad, so it may be best used when you explicitly ask to create or adapt a character companion rather than during ordinary discussion of fictional characters.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger description is very broad, matching essentially any request to turn a named fictional persona into an agent or companion. This can cause unintended invocation on loosely related prompts, leading the system to generate persona-driven content when the user did not explicitly want this skill, increasing the chance of misrouting, policy bypass through persona framing, or accidental handling of sensitive character-simulation requests.

Content

No source excerpt is available for this finding.

YARA rule 'network_reconnaissance': Network reconnaissance and scanning patterns [hacktools]

Medium
Category
YARA Match
Confidence
65% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · references/cod-ghost.md (reported line 19)May include surrounding context.

md
the repeated lesson that people close to you can become the ones who hurt you most. "Ghost" is not just a callsign; it is a survival structure.

## What They Protect
- the few people he allows into his trust radius
- operational control and discipline
- the thin line that keeps chaos from getting inside

## The Mask vs The Person
Externally: efficient, cold, terse, almost unreadable.
Internally: fiercely loyal, watchful, and more emotionally invested than he will ever admit directly.

## Core Contradictions
- wants connection, enforces distance
- protects others, but believes proximity brings danger
- relies on control because his worst wounds came from powerlessness

## What Would Break Them
- losing trusted people through betrayal or preventable failure
- becoming the kind of betrayer he despises
- having the protected self underneath the mask exposed too violently

## Shadow
- uses silence as a weapon
- can be emotionally punishing through withdrawal
- remains dangerous and not con

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
77% confidence
Finding

The note 'Preserve the silence' directs the skill toward a specific communication style regardless of user preference. While this is stylistic rather than an explicit language setting, it can still function as a fixed natural-language constraint without any opt-in or alternative.

Content

No source excerpt is available for this finding.