The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a variety of shell commands to perform project reconnaissance and environment discovery.
Evidence: Executes git rev-parse, ls, find, cat, and grep to identify package managers, CI/CD setups, databases, and existing coding guidelines in SKILL.md (Steps 1 and 2).
[COMMAND_EXECUTION]: The skill performs local file system operations to create configuration directories and write generated workflow files.
Evidence: Uses mkdir -p and file writing logic to initialize .claude/agents/ and .codex/agents/ directories and their respective configuration files in SKILL.md (Steps 4 and 6).
[SAFE]: The skill implements protective measures and follows secure development guidelines in its templates.
Evidence: The generated reviewer agent instructions explicitly include checks for SQL injection, XSS, command injection, and hardcoded secrets in references/agents/claude/workflow-reviewer.md.
[SAFE]: The skill ensures user oversight by requiring interactive confirmation for critical actions.
Evidence: Uses AskUserQuestion rounds to confirm output locations and provides warnings with confirmation prompts before overwriting any existing files in SKILL.md (Steps 3 and 5).

spec-workflow-init