The Agent Skills Directory

[DATA_EXFILTRATION]: The skill correctly handles sensitive information by instructing users to use Harness 'secret' and 'connector' types instead of hardcoding credentials. It promotes the use of environment-specific secret files (e.g., $HARNESS_OUTPUT_SECRET_FILE) for passing sensitive data between pipeline steps.
[COMMAND_EXECUTION]: The skill provides templates that include shell script steps and container-based plugins. These are standard features of the Harness pipeline environment and are documented with security best practices in mind, such as avoiding hardcoded tokens and using secure input types.
[PROMPT_INJECTION]: While the skill itself does not contain malicious prompt injection patterns, the templates it generates (specifically the 'Documentation Generator Agent') include LLM prompts that process external codebases. These generated prompts do not include explicit boundary markers or instructions to ignore embedded commands in the processed data, representing a surface for indirect prompt injection in the resulting agents.

create-agent-template