The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from external targets through web crawling, API probing, and source code auditing. When combined with its significant local capabilities, this creates a vulnerability surface for indirect prompt injection where a malicious target could attempt to influence the agent's behavior via crafted responses or poisoned source code.
Ingestion points: The skill ingests external content using tools like katana, gau, and waybackurls, and audits third-party source code and GitHub Action workflows.
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present for processing tool outputs.
Capability inventory: The skill utilizes full shell execution, file system writes (via -o and anew), and extensive network communication (curl, ffuf, nuclei).
Sanitization: There is no evidence of explicit sanitization for external content before it is presented to the agent's context.
[EXTERNAL_DOWNLOADS]: The skill provides instructions to download and install a wide array of third-party security tools and dependencies from external repositories and registries.
Evidence: The documentation includes commands to install tools such as arjun, paramspider, kiterunner, xsstrike, sqlmap, and semgrep using pip, go install, and brew from various public sources.
[COMMAND_EXECUTION]: The core functionality of the skill involves the agent executing complex shell pipelines and automated reconnaissance tools on the host system.
Evidence: Extensive use of CLI tools including subfinder, httpx, dnsx, nuclei, and ffuf to perform active network probing and scanning.

bug-bounty