The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill requires the FITTINGS_PRIVATE_KEY environment variable for operation. The documentation acknowledges that this sensitive key is stored in plaintext in the execution environment, which poses a risk of credential exposure.
[EXTERNAL_DOWNLOADS]: The SKILL.md file contains instructions to download a script (call.mjs) from fittings.sh using curl. This involves fetching executable code from an external source at runtime to perform service calls.
[DATA_EXFILTRATION]: The call.mjs script accesses the user's private key to generate cryptographic signatures for payment headers sent to the fittings.sh domain. While used for its stated purpose, this involves processing sensitive cryptographic credentials for remote transmission.
[COMMAND_EXECUTION]: The skill relies on an external Node.js script (call.mjs) to manage network communication, handle payment handshakes, and process command-line arguments.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8).
Ingestion points: The records array in the POST request body.
Boundary markers: Absent; there are no instructions to the agent to ignore embedded commands within the input records.
Capability inventory: The call.mjs script performs network requests and cryptographic signing.
Sanitization: Input is processed via JSON parsing, but there is no specific validation of the content of the records or the resulting schema.

schema-infer