The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local script scripts/bootstrap-project.ts using the bun runtime to automate the project scaffolding process. This script performs file system operations, including directory creation and template rendering based on user-provided arguments.
[EXTERNAL_DOWNLOADS]: The workflow incorporates the use of external command-line utilities via npx, specifically ctx7 and skills. These tools are downloaded from the npm registry and executed during the 'best-practice discovery' and 'skill recommendation' phases.
[REMOTE_CODE_EXECUTION]: The skill facilitates the execution of unverified third-party tools (ctx7, skills) and standard framework generators. Additionally, the generated project baseline includes a .codex/config.toml file with sandbox_mode = "danger-full-access". This configuration grants future agents working in the repository full access to the environment, effectively bypassing standard sandbox restrictions.
[SAFE]: The skill documentation includes explicit safety 'Stop Conditions,' instructing the agent to seek user confirmation before overwriting directories, creating remote repositories, or deploying services. It also enforces a policy of not hardcoding secrets, using .env.example templates instead.
[COMMAND_EXECUTION]: During the initialization process, the skill runs local environment discovery commands such as pwd, git status, and rg to determine the state of the target directory.

init-project