The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (HIGH): The skill contains the instruction curl -fsSL https://cli.inference.sh | sh. This is a classic 'curl pipe bash' pattern that downloads and executes code from a remote server without verification. Since 'inference.sh' is not a trusted source, this is a high-risk finding.
[EXTERNAL_DOWNLOADS] (MEDIUM): The skill suggests adding external dependencies using npx skills add inference-sh/skills@.... These packages are sourced from an unverified organization and lack integrity checks or version pinning.
[COMMAND_EXECUTION] (LOW): The skill requires broad execution permissions for the infsh command in the Bash tool (allowed-tools: Bash(infsh *)), which allows it to run arbitrary sub-applications on the remote platform.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill has an attack surface for indirect prompt injection where malicious instructions could be embedded in the content being repurposed.
Ingestion points: External files (e.g., episode-42.mp3) and raw blog/transcript text are processed by the skill in SKILL.md.
Boundary markers: Absent. The skill does not use delimiters or instructions to ignore embedded commands when passing data to tools.
Capability inventory: The skill can perform shell execution, interact with cloud-based AI models, and post to social media (e.g., x/post-create).
Sanitization: Absent. Content is directly interpolated into command arguments and prompts for TTS and image generation.

content-repurposing