The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute several local Python scripts (manage_style_registry.py, style_fingerprint.py, normalize_style_frontmatter.py, and sync_claude_runtime.py) to manage its internal data and analyze text samples. The scripts provided in the package demonstrate high security maturity, including strict path validation to prevent directory traversal and the use of temporary files for atomic data persistence. Note that sync_claude_runtime.py is referenced in the logic but was not included in the analyzed files.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process external article content fetched from URLs, which presents a surface for indirect prompt injection attacks.
Ingestion points: External articles are fetched via the web-article-extractor tool and stored in docs/*.md before being parsed by analysis scripts and the agent.
Boundary markers: The skill lacks explicit boundary markers or instructions to ignore embedded commands within the analyzed text, which could allow malicious content in a processed article to influence the agent's behavior during the modeling process.
Capability inventory: The skill possesses capabilities to execute shell commands (running local scripts) and perform filesystem writes (updating the style registry and generating markdown profiles).
Sanitization: The text analysis script (style_fingerprint.py) focuses on markdown structure and punctuation statistics; it does not sanitize the text for potential prompt injection patterns.

style-modeler