The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a variety of shell commands, including docker run, uv run, and bash. These commands are used to process images, generate scripts, and run analysis tools located at specific absolute paths (e.g., /Users/mtm/pdev/taylormonacelli/). There is a risk of command injection if parameters like the feed time (FEED) are not properly sanitized before being interpolated into shell commands.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources, specifically time-lapse photos and text manifests generated by other scripts, which could potentially contain malicious content intended to influence the agent's behavior.
Ingestion points: The agent reads image metadata from /Users/mtm/dcim2/OpenCamera and parses text output from SCRATCH/surface.txt and SCRATCH/frames-list.txt (SKILL.md).
Boundary markers: There are no explicit delimiters or instructions for the agent to ignore potentially embedded instructions within the processed files.
Capability inventory: The skill possesses significant capabilities, including the ability to run Docker containers with volume mounts and execute arbitrary Python scripts via uv run (SKILL.md, list_frames.py, surface_height.py).
Sanitization: The skill lacks evidence of input sanitization or validation for the data ingested from the file system.
[EXTERNAL_DOWNLOADS]: The skill utilizes the linuxserver/ffmpeg Docker image to perform video and image processing tasks. This image is sourced from LinuxServer.io, a well-known community-maintained repository.

starter-peak-from-photos