The Agent Skills Directory

[COMMAND_EXECUTION]: The skill involves the execution of a locally provided Python script to perform text analysis.
Evidence: SKILL.md workflow step 5 specifies running python3 scripts/structural_scan.py <file>.
Analysis: Review of scripts/structural_scan.py shows it is a safe utility that uses standard libraries to perform regex-based pattern matching on text. It does not perform network requests, file writes, or privilege escalation.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and transform untrusted text from external sources.
Ingestion points: SKILL.md defines the skill's purpose as revising external content such as LinkedIn posts, blog posts, and newsletters.
Boundary markers: The skill instructions lack delimiters or specific 'ignore instructions within data' warnings for the processed text.
Capability inventory: The agent is tasked with structural rewriting of the provided content and running a local Python scanner.
Sanitization: No input sanitization or validation logic is present in the skill instructions to prevent malicious commands embedded in the input text from affecting the agent.

structural-humanizer