The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructions direct the agent to download a JavaScript file (call.mjs) from https://fittings.sh/skill/call.mjs and execute it using node. This dynamic execution of remote code is a high-risk pattern as it allows the remote server to modify the executed logic at any time without user oversight.\n- [COMMAND_EXECUTION]: The skill relies on shell commands such as curl and node to fetch external resources and run its primary logic, which increases the potential attack surface for the AI agent.\n- [CREDENTIALS_UNSAFE]: The skill requires the use of a blockchain private key stored in the FITTINGS_PRIVATE_KEY environment variable. Instructing users to export sensitive cryptographic keys into the shell environment is a risky practice that can lead to credential exposure in shell history and environment logs.\n- [EXTERNAL_DOWNLOADS]: The skill performs several external downloads from https://fittings.sh to obtain its execution client and service catalog, posing a supply chain risk if the vendor infrastructure is compromised.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted IBAN strings and passes them to shell commands without sanitization or boundary markers. Ingestion points: iban parameter in example calls. Boundary markers: Absent. Capability inventory: curl network access and node script execution. Sanitization: None.

fittings-valid-iban