The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to download updates from the author's GitHub repository.
Evidence: Found in the "Skill version handshake" section: https://github.com/chongdashu/spriterrific-skills/releases/latest/download/spriterrific-api-skill.zip.
[DYNAMIC_EXECUTION]: The update procedure involves unzipping a remote archive directly into the project root, which overwrites existing skill definitions.
Evidence: Instructions state to "unzip it at the project root (it refreshes both .claude/skills/ and .agents/skills/ copies)". This is a form of self-modifying code execution.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data which could potentially contain malicious instructions meant for the sprite-generation engine or the agent.
Ingestion points: The sourceImageUrl parameter (remote image URL) and editPrompt / actionContext fields (user-controlled text).
Boundary markers: No delimiters or explicit "ignore embedded instructions" warnings are defined for these inputs.
Capability inventory: The skill uses curl for network operations and writing files to the local file system.
Sanitization: No sanitization or validation of the input strings is mentioned before they are sent to the API.
[COMMAND_EXECUTION]: The skill relies on shell commands for API interaction and file management.
Evidence: Multiple examples using curl to send POST/GET requests and save artifacts locally using -o flags.

spriterrific-api