The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill is designed to capture and store sensitive target information, including 'authentication details' and 'Testing accounts: Credentials for authenticated testing', within local markdown files (targets/{target}.md). While this is functional for the intended use case of bug bounty hunting, it involves the handling and storage of plaintext credentials within the agent's file system.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection attack surface by processing untrusted external data. It specifically instructs the agent to analyze JavaScript bundles, source code intelligence from GitHub, and decompiled mobile app strings.
Ingestion points: analyze-javascript.md, analyze-mobile.md, and recon-passive.md (reading external JS and source code).
Boundary markers: No explicit instruction for delimiters or 'ignore embedded instructions' warnings are provided when processing this external content.
Capability inventory: The skill possesses significant capabilities including file read/write operations and the execution of various CLI tools via the Task tool.
Sanitization: There is no evidence of sanitization or filtering of the external content before it is processed by the agent.
[COMMAND_EXECUTION]: The skill methodology relies heavily on the execution of external CLI tools such as subfinder, amass, httpx, dnsx, gospider, ffuf, apktool, and truffleHog. It provides 'exact commands' for the agent to follow, which assumes these tools are available and trusted in the execution environment.

bug-bounty-methodology