The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to clone a repository from a third-party account (github.com/doane2002cn/moontv-openclaw-skill.git) and to install multiple external Python dependencies including 'requests' and 'openai'.
[REMOTE_CODE_EXECUTION]: The skill requires the execution of Python scripts (moontv_daily.py) that are downloaded from an unverified external source, which allows for arbitrary code execution in the agent's environment.
[COMMAND_EXECUTION]: The skill documentation includes shell commands for environment configuration and script execution (git clone, cp, python) that the agent is expected to perform.
[DATA_EXFILTRATION]: The skill performs network operations to non-standard domains, including a movie data gateway at 'moontv-api.12879737.xyz' and a third-party LLM proxy at 'api.gptgod.online'.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection. It fetches movie titles and synopses from third-party CMS APIs and interpolates them directly into LLM prompts without sanitization.
Ingestion points: External movie data is fetched via 'requests.get' from the MoonTV gateway and various CMS source APIs in 'moontv_daily.py'.
Boundary markers: There are no markers or safety instructions in the 'build_prompt' function to prevent the LLM from following instructions embedded in the movie synopses.
Capability inventory: The skill has capabilities for network access and file system writes (generating reports and cleaning up files).
Sanitization: No validation or escaping is applied to the untrusted media content before it is processed by the LLM.

moontv-openclaw-skill