The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from creator bios, hashtags, and link-in-bio pages through various search tools (e.g., v1_instagram_search_profiles, v1_linktree).
Ingestion points: Search results and profile data are collected in scripts/collect_pool.py, scripts/enrich.py, and scripts/contacts.py.
Boundary markers: The instructions do not define explicit boundary markers or warnings for the agent to ignore instructions embedded within the sourced data.
Capability inventory: The skill has the capability to execute shell commands (via Bash), write files, and publish artifacts.
Sanitization: While scripts/build_dashboard.py includes HTML escaping (esc() function) to prevent XSS in the generated dashboard, there is no specific sanitization to prevent the agent itself from following instructions found in the bio text during the reporting phase.

creator-sourcing