The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill features capabilities that ingest untrusted data from external sources, which could potentially be used for indirect prompt injection attacks.
Ingestion points: The /ads dna <url> command fetches brand information from external websites. The /ads update command retrieves data from community platforms like Reddit and Hacker News to update internal reference files used by the agents.
Boundary markers: There are no documented boundary markers or instructions to disregard embedded commands in the processed data.
Capability inventory: The skill invokes multiple Python scripts and interacts with various external APIs for data retrieval and image generation.
Sanitization: No sanitization or validation procedures are described for the fetched content before it is integrated into the agent's context.
[COMMAND_EXECUTION]: The skill relies on the execution of several local Python scripts for core operations, including scripts/profile.py for context management and scripts/api/meta_fetch.py for platform data integration.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for installing community-maintained packages such as mcp-google-ads and tiktok-ads-mcp-server via standard package managers. It also fetches updates from well-known services including the Reddit API, Hacker News, and official ad platform changelogs to refresh its reference materials.

ads