The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute multiple local Python scripts (e.g., readability-analyzer.py, ai-tell-scan.py, run-audit.py) located in ${CLAUDE_PLUGIN_ROOT}/scripts/ to automate content quality assessments and authorship verification. These scripts use parameters derived from brand configurations and generated file paths.\n- [INDIRECT_PROMPT_INJECTION]: The skill provides a mechanism for ingesting untrusted external content via the --source-draft parameter. While it employs an extensive quality pipeline to filter and verify generated content, the ingestion of raw user drafts represents a surface for indirect prompt injection.\n
Ingestion points: Rough draft text files (00-source-draft.md) and content topics provided in the initial user prompt.\n
Boundary markers: The skill uses a numbered file convention and a sequential gated pipeline to separate research, drafting, and verification phases.\n
Capability inventory: The skill can read local brand profile JSON files, write content artifacts to the file system, and execute specialized local Python tools via subprocess calls.\n
Sanitization: Content is reviewed through a 'Humanize' scan to detect AI patterns, a brand-voice scorecard for stylistic alignment, and a fact-checking gate to verify claims against citations.

content-engine