The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from multiple sources including client-provided materials (pitch decks, audit outputs, brand-voice documents), kickoff transcripts, and founder notes stored in ~/marketing-plans/{client-slug}/materials/. This content is processed and interpolated into generated marketing plans and roadmaps.
Ingestion points: The skill reads files from the materials/ directory and pulls data from external APIs and transcripts during the INIT phase.
Boundary markers: No explicit boundary markers or "ignore embedded instructions" warnings are defined for the processing of these external materials.
Capability inventory: The skill has capabilities to read/write local files, access external APIs (Stripe, Google Analytics 4, Ahrefs, Typefully) via MCP integrations, and publish content to GitHub repositories.
Sanitization: The instructions do not specify sanitization or validation of the ingested external content before it is processed by the agent.
[DATA_EXFILTRATION]: The skill is designed to pull sensitive business and financial data from services like Stripe (MRR, ARR, churn, LTV) and Google Analytics 4. It also supports publishing the final compiled plan to a shared GitHub repository. While this is a core functional requirement, users should monitor data flows to external repositories. The skill includes a verification pass during the FINALIZE phase to check for machine-specific paths before shipping.
[EXTERNAL_DOWNLOADS]: The skill references several external CLI tools and well-known services for its operations, including agent-browser, defuddle, obsidian-cli, and various MCP integrations such as Ahrefs, Google Analytics 4, and Typefully. These are documented as standard tools for research, analytics, and social media management.

marketing-plan