The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes a local script named language-router.py using CLI arguments in several steps of its process (Steps 4, 5, and 7) to validate supported languages and determine translation routing. This represents the execution of logic outside the skill's primary instruction set.
[DYNAMIC_EXECUTION]: The agent performs dynamic invocation of the language-router.py utility with arguments constructed at runtime based on brand slugs and language codes derived from configuration files.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it processes data from several external files that could contain adversarial instructions.
Ingestion points: Reads from brand guidelines (_manifest.json), the brand profile (profile.json), and agency Standard Operating Procedures (SOPs) located in ~/.claude-marketing/sops/.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions to prevent the agent from obeying commands embedded within these data sources.
Capability inventory: The skill has file system read/write access to the profile.json and the ability to execute the language-router.py script.
Sanitization: No sanitization or filtering of natural language content is performed on the ingested files beyond basic JSON integrity checks.
[DATA_EXFILTRATION]: The skill accesses the ~/.mcp.json file to verify the presence of configured Model Context Protocol servers. While used for validation, this provides access to environment-level configuration data that may contain sensitive server paths or metadata.

language-config