The Agent Skills Directory

[SAFE]: The skill correctly manages sensitive credentials by requiring Meta Marketing API access tokens to be stored in a .env file rather than hardcoding them in scripts. Reference: scripts/config.py.
[SAFE]: Network operations are restricted to well-known, official service endpoints for the Meta Graph API (graph.facebook.com). No suspicious or unknown external domains are contacted. Reference: scripts/client.py.
[SAFE]: The skill implements privacy-preserving measures for audience creation by SHA256-hashing personally identifiable information (PII) before transmission to the API. Reference: scripts/create_custom_audience.py.
[SAFE]: Persistence for the ad scheduling feature is handled via a local JSON configuration file within the skill directory, and the suggested automation via cron is a standard administrative practice for its intended function. Reference: scripts/ad_scheduler.py.
[SAFE]: No obfuscation, privilege escalation, or dynamic execution patterns were detected across the skill's scripts and knowledge base files.
[SAFE]: Indirect prompt injection surface exists as the skill processes ad creative data from the Meta API, but this is a standard operational requirement with no detected malicious intent.
Ingestion points: scripts/get_active_ads_copy.py, scripts/list_ads.py
Boundary markers: Not explicitly defined in instructions
Capability inventory: Status and budget modification via scripts/update_campaign.py and scripts/ad_scheduler.py
Sanitization: Standard API JSON parsing; no natural language sanitization defined

meta-ads