The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts (scripts/language-router.py, scripts/eval-runner.py, and scripts/brand-voice-scorer.py) to perform tasks such as translation routing, content scoring, and brand voice assessment. These scripts take parameters like market codes, content types, and localized text as command-line arguments.
[DATA_EXPOSURE]: The skill accesses project-specific data and configuration files located in the home directory under ~/.claude-marketing/. This includes brand profiles, guidelines, and active status files used to maintain context across sessions.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect injection surface as it ingests and processes untrusted external content (campaign assets like emails, ads, and landing pages).
Ingestion points: Campaign assets provided via file paths, directories, or inline content blocks (Process Step 2).
Boundary markers: No specific delimiters or instructions to ignore embedded commands are mentioned in the prompt templates.
Capability inventory: The skill can execute subprocesses via the mentioned Python scripts and call external translation MCP servers (deepl, sarvam-ai, etc.).
Sanitization: There is no evidence of sanitization or escaping of the ingested campaign content before it is passed as arguments to scripts or processed for transcreation.

localize-campaign