The Agent Skills Directory

[SAFE]: The skill provides strategic frameworks and playbooks for scaling SEO-driven pages. The instructions emphasize unique value and quality, adhering to search engine guidelines and user intent. No prompt injection or bypass attempts were found.- [COMMAND_EXECUTION]: The file scripts/url_pattern_generator.py is a utility for generating URL permutations based on user-provided templates and variables. It uses standard Python libraries (json, sys, os, itertools) to perform string manipulation. It does not use eval(), exec(), or subprocess to run arbitrary commands.- [DATA_EXPOSURE]: The skill instructions suggest reading a local context file (.claude/product-marketing-context.md) if it exists. This is a standard pattern for personalizing agent output based on project context and does not involve unauthorized access or network exfiltration.- [INDIRECT_PROMPT_INJECTION]: The skill processes external data through both the prompt context and the URL generation script. (1) Ingestion point: .claude/product-marketing-context.md and user-provided JSON files for the script. (2) Boundary markers: Not explicitly defined in the prompt templates. (3) Capability inventory: The script can read files via open() but is limited to JSON parsing and string replacement. (4) Sanitization: The script performs basic sanitization on slug values (lowercase and hyphen replacement). The risk of indirect injection is negligible as the processed data is used for structural planning rather than autonomous execution.

programmatic-seo