Security audit

story-spark

Security checks across malware telemetry and agentic risk

Overview

This skill does what it says: it locally turns user-selected photos and text files into writing prompts, but users should treat the inputs and exported prompts as private.

Install only if you are comfortable letting the tool read the specific photo or journal folders you point it at. Prefer small curated folders, avoid files with highly sensitive third-party or location data, and review exported JSON before sharing, syncing, or committing it.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
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 (10)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 95% confidence
Finding: The skill instructs users to read from personal photo and journal directories and to export generated results, which implies file read/write behavior, but the skill does not declare permissions or clearly bound what data may be accessed. This creates a transparency and consent gap: a user or hosting platform may not realize the skill can process sensitive local content, increasing the chance of overbroad access and accidental exposure.

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: The script extracts and stores precise GPS coordinates from photo EXIF into the generated prompt data even though exact latitude/longitude is not necessary to create creative writing seeds. This can expose highly sensitive location history in terminal output or saved JSON, increasing privacy risk if prompts are shared, logged, or retained.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The README encourages scanning personal photos, journal entries, and notes but does not warn users that these are highly sensitive data sources that may contain intimate personal information, identities, relationships, health details, and other confidential content. In a skill explicitly designed to mine emotionally resonant moments from private material, the absence of privacy disclosures and safe-handling guidance increases the risk of users exposing sensitive data without informed consent.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The README explicitly states that the tool reads EXIF metadata including GPS location and timestamps, but it does not warn that this can reveal home addresses, routines, travel history, workplaces, or other sensitive location patterns. Because the skill transforms real-life moments into prompts, location details may also be surfaced in outputs, increasing the chance of inadvertent disclosure or deanonymization.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The description says the skill scans personal photos metadata, journal entries, and notes, but it does not prominently warn that these sources can contain highly sensitive information such as locations, dates, private thoughts, and relationship details. In this context, the omission is risky because the entire workflow is built around mining intimate personal data for creative reuse.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The export feature is presented as a normal workflow step without warning that generated prompts may preserve or infer identifying details from the original photos, EXIF metadata, journals, or notes. This can leak sensitive information into saved text/JSON outputs that are easier to share, sync, or commit elsewhere than the original private sources.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The document explicitly encourages use of personal photos and journal entries, which are highly privacy-sensitive sources, but provides no warning about sensitive content handling, minimization, consent, or retention. In the context of this skill, that omission increases the risk that users expose intimate metadata, locations, timestamps, or private thoughts without understanding the privacy implications.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The skill is explicitly designed to scan personal photos and process EXIF metadata, but it provides no clear warning at startup or before execution that sensitive metadata such as timestamps and location may be read. In a personal-content skill, that omission increases the chance users expose private data without informed consent.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The tool reads journal and text files that may contain intimate personal content, yet it does not warn users that excerpts may be parsed, displayed, and transformed into output prompts. Because the skill's purpose is to mine emotionally resonant personal writing, the privacy sensitivity is higher than for ordinary text processing.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: When --output is used, the script writes generated prompts to disk including source details and, for journal-derived prompts, premise text based on personal excerpts. Persisting this information without explicit disclosure or minimization can create unintended local data retention and secondary exposure risk.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.