The Agent Skills Directory

[SAFE]: The skill follows its stated purpose of prose synthesis and does not exhibit any malicious behaviors such as credential exfiltration, remote code execution, or privilege escalation. All operations are confined to standard text processing and file management.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it processes untrusted content (user prose or file data) and has the capability to write to the file system. This is a characteristic of the skill's intended functionality.
Ingestion points: Reads unstructured thinking from the $ARGUMENTS parameter or through the Read tool for specific file paths.
Boundary markers: The instructions do not use specific delimiters or protective framing to distinguish between the synthesis instructions and the data being processed.
Capability inventory: The skill uses Read, Edit, Write, Grep, Glob, and TodoWrite tools to analyze input and record the synthesized plans.
Sanitization: There is no explicit sanitization or filtering of the input prose before the synthesis steps are performed by the model.

prose-synthesize