The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to clone a third-party repository from https://github.com/ticarpi/jwt_tool and install dependencies using pip3 install -r requirements.txt. These resources originate from an external repository not associated with a pre-approved trusted organization.
[REMOTE_CODE_EXECUTION]: The skill provides instructions to execute the downloaded Python scripts (jwt_tool.py) with various flags. This represents a pattern where code is fetched from an external source and executed locally.
[DATA_EXFILTRATION]: The toolkit includes functionality to send forged tokens to external target URLs using the -t flag, combined with custom headers (-rh) and cookies (-rc). While documented for testing purposes, this represents a network operation capability to external domains.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external, untrusted data in the form of JWT tokens, creating an attack surface for indirect prompt injection.
Ingestion points: The agent accepts user-provided JWT tokens as the primary input for analysis and exploitation scripts.
Boundary markers: There are no explicit instructions or delimiters defined to ensure the agent treats the content of the token as data rather than instructions.
Capability inventory: The skill has the ability to execute shell commands (via python3), perform network requests (via the -t flag), and interact with the local file system (reading keys and wordlists).
Sanitization: The instructions do not specify any validation or sanitization of the token string before it is interpolated into shell commands.

jwt-tool