The Agent Skills Directory

[COMMAND_EXECUTION] (HIGH): The skill relies on multiple external Python scripts (whatsapp_search.py, nas_secrets.py) that are not included in the provided source code. Executing these scripts via Bash tools presents a high risk as their internal logic is unverified.
[CREDENTIALS_UNSAFE] (HIGH): The documentation explicitly instructs the user to export GH_TOKEN to the environment and pass sensitive tokens as command-line arguments to nas_secrets.py. This is a dangerous pattern that can lead to credential exposure in shell history, process listings, or logs.
[DATA_EXFILTRATION] (MEDIUM): The skill is designed to collect private messages and emails into markdown files and then 'push' them to a remote Git repository. While this aligns with the stated purpose of 'intelligence gathering,' it automates the exfiltration of potentially sensitive personal data to the cloud.
[INDIRECT_PROMPT_INJECTION] (LOW): The scripts/gather_intel.py file ingests untrusted data from WhatsApp and news sources to build a candidate profile.
Ingestion points: add_whatsapp_intel and add_news_intel in scripts/gather_intel.py.
Boundary markers: None; the data is directly appended to JSON/Markdown structures.
Capability inventory: File writing (Markdown/JSON), subprocess execution for Git and distribution scripts.
Sanitization: No sanitization or escaping of external content is performed before storage or report generation.

community-campaign-builder