The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of local shell commands to orchestrate security tools.
Evidence includes usage of amass, subfinder, assetfinder, findomain, dnsgen, httprobe, httpx, massdns, whatweb, nuclei, ffuf, waybackurls, gau, unfurl, paramspider.py, Gxss, dalfox, and qsreplace.
A custom bash script recon.sh is provided to automate the subdomain enumeration and vulnerability scanning pipeline.
[EXTERNAL_DOWNLOADS]: The skill interacts with external web services to gather reconnaissance data.
Fetches ASN information from bgp.he.net via curl.
Retrieves historical URL data from services via waybackurls and gau.
These are standard reconnaissance practices using well-known services and do not involve downloading or executing untrusted executable code.
[INDIRECT_PROMPT_INJECTION]: The skill has a potential attack surface for indirect prompt injection by processing data from external sources.
Ingestion points: Reads data from external URLs (via waybackurls, gau, curl) and saves them to local files (e.g., urls.txt, live.txt) which are subsequently processed by other tools.
Boundary markers: The skill explicitly includes a "Mandatory confirmation gate" requiring the agent to ask for target URLs, scope, and command confirmation before proceeding.
Capability inventory: Extensive command execution capabilities across all recon steps in SKILL.md.
Sanitization: The skill does not explicitly describe sanitization of the external data before it is processed by subsequent tools or the agent's context.

red-team-tools