The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The documentation and references identify dependencies on several standard Python libraries, including requests, pandas, jinja2, and PyYAML. These are used for network communication, data processing, and template management and are expected to be installed from official package registries.
[COMMAND_EXECUTION]: SKILL.md provides instructions for setting up a crontab entry to execute the processing script hourly. This is a standard persistence and automation method for monitoring tools and is consistent with the skill's stated purpose of providing automated alerting.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection within scripts/process.py which could be exploited if an attacker provides a malicious findings CSV.
Ingestion points: The import_findings function in scripts/process.py reads vulnerability data, including CVE IDs and hostnames, directly from a user-supplied CSV file.
Boundary markers: The script does not utilize any delimiters or instructions to treat the interpolated data as untrusted when generating alerts.
Capability inventory: The script has the capability to send notifications to Slack, PagerDuty, and via SMTP email, as well as writing to a local SQLite database.
Sanitization: Fields such as cve_id and asset_hostname are interpolated directly into notification messages without escaping or sanitization, allowing for potential manipulation of the alert text seen by security responders.

implementing-vulnerability-sla-breach-alerting