The Agent Skills Directory

[SAFE]: The skill is purely instructional and contains no executable code, malicious patterns, or dangerous commands. It serves as a developer guide for building integration tools.
[COMMAND_EXECUTION]: The instructions reference the use of command-line tools such as agent-eval-harness, grill-me, and tdd. These are described as local development utilities for schema discovery and contract testing.
[INDIRECT_PROMPT_INJECTION]: The skill defines a specification for processing external JSON data via stdin. While any system processing external data for an agent has an input surface, these instructions emphasize strict schema validation and contract testing, which are security best practices.

trial-adapters