The Agent Skills Directory

[PERSISTENCE]: The skill modifies shell startup configuration files (~/.zshrc) to persist environment variables like the Chaos API key. It also provides explicit instructions and scripts to be scheduled via cron for continuous monitoring of subdomains and GitHub commit activity.- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process large volumes of untrusted data from the internet, creating an attack surface where malicious data could influence the agent's behavior.
Ingestion points: Data is fetched from certificate transparency logs (crt.sh), DNS discovery services (ProjectDiscovery), GitHub APIs, and via direct crawling of target web assets using tools like katana and waybackurls.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present in the scripts that process external data.
Capability inventory: The skill possesses extensive shell command execution capabilities, file system write access to /tmp and local directories, and network access for reconnaissance.
Sanitization: Sanitization is limited to standard line processing (jq, grep, sort, uniq) which does not prevent specialized injection payloads from being processed by downstream tools.- [COMMAND_EXECUTION]: The skill executes a wide range of shell commands to manage external binaries and local files. It writes configuration files for subfinder to the user's home directory (~/.config/subfinder/config.yaml) and manages Python virtual environments for tools like SecretFinder and LinkFinder.- [EXTERNAL_DOWNLOADS]: Fetches data and configuration from well-known security services and public research platforms, including GitHub, crt.sh, and ProjectDiscovery's infrastructure. It also includes commands to install the trufflehog3 package from PyPI.

web2-recon