The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute multiple local shell commands for media processing and project maintenance.
Evidence: Commands include ffmpeg for frame extraction, sips for image compression, and pnpm for running type checks and linters (pnpm types:check, pnpm lint). These are standard operations for the described developer workflow.
[DATA_EXFILTRATION]: The skill automates the transfer of local image assets to remote storage.
Evidence: The workflow uses wrangler r2 object put to upload files to the 01mvp-public-assets bucket and uses curl -I to verify availability on assets.01mvp.com. These domains are identified as vendor resources associated with the author 'makerjackie'.
[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection via the ingestion of untrusted subtitle data.
Ingestion points: The agent reads and processes external .srt files as part of Step 2 in the workflow defined in SKILL.md.
Boundary markers: None identified. There are no explicit instructions to the agent to treat the subtitle text as untrusted or to ignore embedded instructions within those files.
Capability inventory: The agent has access to shell execution (pnpm, ffmpeg), network write capabilities (wrangler), and repository modification (git).
Sanitization: The instructions focus on correcting transcription errors for stylistic consistency but do not provide mechanisms for sanitizing or escaping potentially malicious instructions within the subtitles.

mj-video2md-html