The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill includes a self-update utility in scripts/update-skill.sh that downloads executable script files from a remote GitHub repository, replaces local files, and applies execution permissions via chmod +x. This allows for arbitrary code execution on the agent's host system if the remote source is compromised.
[COMMAND_EXECUTION]: The skill relies on shell command execution via bash scripts (scripts/genviral.sh) to perform its core functions. It also provides instructions for setting up persistence using openclaw cron jobs, which execute these scripts automatically in the background without direct user oversight.
[EXTERNAL_DOWNLOADS]: The skill connects to non-whitelisted external domains to fetch data and updates. While it primarily targets the Genviral API and GitHub, these connections occur outside the predefined safety boundaries for network operations.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from external social media platforms (TikTok) using a browser tool during research and trend analysis tasks.
Ingestion points: Social media search results and creator profiles processed via docs/references/competitor-research.md and docs/api/analytics.md.
Boundary markers: Absent; there are no delimiters or instructions to ignore embedded commands in the processed data.
Capability inventory: Extensive subprocess execution capabilities via scripts/genviral.sh and the ability to update its own executable code.
Sanitization: Absent; the skill does not implement validation or escaping for the external content it processes.

genviral