The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/repo-rename.py executes shell commands using subprocess.run() to interact with Git and the GitHub CLI (gh). It calls commands like git rev-parse, git status, and gh repo rename. These are executed using argument lists, which prevents shell injection, and the target name is validated for directory separators and hidden prefixes.- [DATA_EXPOSURE]: The skill accesses and modifies configuration and session files located in private application directories, specifically ~/.claude/projects, ~/.codex/sessions, and ~/.codex/config.toml. While these operations are documented as necessary for maintaining AI agent continuity after a rename, they involve reading and writing to sensitive project metadata and transcript stores.- [INDIRECT_PROMPT_INJECTION]: The skill performs bulk literal byte replacements within repository files and agent configuration files based on the renaming plan. This constitutes a surface for processing potentially untrusted data that enters the agent's context through the file system. \n
Ingestion points: Files within the current repository and agent-specific configuration directories (~/.claude/projects, ~/.codex/sessions, ~/.codex/config.toml). \n
Boundary markers: None; the script performs deterministic byte replacement without content sanitization beyond name validation. \n
Capability inventory: File system read/write, directory manipulation, and external command execution via Git and GitHub CLI. \n
Sanitization: The new_name parameter is validated to disallow directory separators (/) and leading dots (.), mitigating basic path traversal risks.

repo-rename