The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external Framer sites and CMS exports which could potentially contain malicious instructions.
Ingestion points: framer-csv-import.mjs ingests CSV files and API JSON dumps containing rich text (HTML).
Boundary markers: The instructions do not define specific prompt delimiters, but the scripts implement robust programmatic sanitization.
Capability inventory: The skill can write files to the local content/ directory and perform network requests to the Framer API.
Sanitization: The library scripts/lib/content-common.mjs contains an allowlist-based HTML sanitizer (sanitizeHtml) that explicitly drops dangerous tags like <script>, <style>, and event handlers.
[DYNAMIC_EXECUTION]: The skill uses dynamic loading to interface with project-specific dependencies.
Evidence: scripts/framer-api-dump.mjs uses createRequire and dynamic import() to load the framer-api package from the local node_modules. This is a standard pattern for CLI tools to support user-installed peer dependencies for a specific project.

migrate-framer