The Agent Skills Directory

[SAFE]: The skill primarily serves as an educational and technical resource for context engineering, providing guidelines to maximize reasoning performance while minimizing token overhead.
[COMMAND_EXECUTION]: The test suite in scripts/tests/test_edge_cases.py utilizes subprocess.run to execute the skill's own diagnostic scripts. These executions are performed using structured argument lists and are confined to validating internal logic within a controlled test environment.
[EXTERNAL_DOWNLOADS]: Documentation references the official Anthropic Usage Limits API (https://api.anthropic.com/api/oauth/usage) for monitoring service quotas. This interaction involves a well-known service and is consistent with the skill's purpose of runtime awareness.
[CREDENTIALS_UNSAFE]: The skill describes the default file system locations for Claude authentication credentials (e.g., ~/.claude/.credentials.json) to explain how the agent authenticates with the model provider's official API. This is documented for configuration purposes and does not contain hardcoded secrets or malicious harvesting code.

context-engineering