The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, such as user-pasted text and local image files, and passes them to a local analysis tool. This creates a surface for indirect prompt injection where malicious instructions embedded in the analyzed data could theoretically attempt to influence the agent's subsequent actions.
Ingestion points: User-provided text and file paths read via the Read tool in SKILL.md Step 1.
Boundary markers: The skill specifies structured markdown blocks for outputting verdicts and cleaned text in SKILL.md Step 4.
Capability inventory: The skill uses the Bash tool to execute the slop CLI binary for all analysis operations.
Sanitization: The skill includes a dedicated cleanup operation designed to remove zero-width characters and homoglyphs from text.
[COMMAND_EXECUTION]: The skill invokes a local command-line interface located at /Applications/Slop Or Not.app/Contents/MacOS/slop using the Bash tool. This is the intended mechanism for performing on-device AI detection and readability scoring without uploading data to the cloud.

slop-check