The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructs users to store a blockchain private key in the FITTINGS_PRIVATE_KEY environment variable. The instructions explicitly state that the key sits in plaintext wherever the agent runs, creating a high risk of credential exposure or theft if the environment is accessed by unauthorized processes or other AI instructions.
[REMOTE_CODE_EXECUTION]: The skill facilitates remote code execution by instructing the agent to download a script from a remote URL (curl -sO https://fittings.sh/skill/call.mjs) and execute it using node. This pattern allows for the execution of arbitrary code that is not part of the skill's static distribution.
[EXTERNAL_DOWNLOADS]: The skill performs runtime downloads of configuration metadata and execution scripts from the vendor's domain (fittings.sh).
[COMMAND_EXECUTION]: The skill relies on shell command execution, including curl for networking, jq for data processing, and node for executing downloaded scripts, which provides a significant capability surface.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from the National Weather Service via a proxy. As this data is external and interpolated into the agent's context, it presents an attack surface for indirect prompt injection.
Ingestion points: API response from https://fittings.sh/v1/nws/active-alerts (originating from NOAA).
Boundary markers: None identified in the provided instructions.
Capability inventory: Shell command execution (curl, node), environment variable access, network requests, and file writes.
Sanitization: No sanitization or validation of the NWS alert content is mentioned before it is processed by the agent.

fittings-nws-active-alerts