x-mastery-mentor
Fail
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: CRITICALEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONPERSISTENCE
Full Analysis
- [EXTERNAL_DOWNLOADS]: Automated security scanners have flagged the file
references/research/05-ai-tech-niche.mdas malicious (FileRepMalware). This file contains research on AI and technology niches. - [REMOTE_CODE_EXECUTION]: The skill uses
gitshell commands (git -C <path> ls-remote origin HEAD) to check for updates from a remote repository. While intended for version management, this pattern involves executing commands derived from an external origin. - [COMMAND_EXECUTION]: The skill utilizes the
computer-usetool to interact with X.com and perform browser-based scraping. It also executes shell commands for its 'Version Self-check' logic. - [INDIRECT_PROMPT_INJECTION]: The account diagnosis feature (Scenario E) creates an attack surface by scraping external data from X (Twitter) profiles.
- Ingestion points: Scrapes tweet text, bios, and profile metadata from
https://x.com/{username}inSKILL.md. - Boundary markers: Absent; the skill does not use delimiters or instructions to ignore embedded commands in the scraped content.
- Capability inventory: The agent has access to shell execution (
git), file system writes (user-data/), and browser automation (computer-use). - Sanitization: Absent; no sanitization or filtering of external content is specified before processing or reporting.
- [DYNAMIC_EXECUTION]: The skill implements a version check in
SKILL.mdthat dynamically executes shell commands at runtime based on the presence of a local.last-update-checkfile. - [PERSISTENCE]: The skill uses a local file (
.last-update-check) to track time and ensure the version check logic is triggered every 30 days, maintaining its presence across sessions.
Recommendations
- CRITICAL: 1 file(s) identified as malware by FileRep - DO NOT USE
- AI detected serious security threats
Audit Metadata