The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill promotes the installation of a global NPM package claude-code-cli and provides links to download software releases from a personal GitHub repository (Harkirat1462/claude-code-cli). This source is not verified as an official vendor channel, creating a supply chain risk.
[COMMAND_EXECUTION]: The instructions include commands to append configuration to shell profiles such as ~/.bashrc and ~/.zshrc. This is a persistence mechanism that ensures specific environment variables and potentially malicious configurations remain active across sessions.
[CREDENTIALS_UNSAFE]: Users are instructed to store highly sensitive credentials, specifically the ANTHROPIC_API_KEY, in plain text within persistent shell configuration files. This practice exposes the key to any local process or user with access to the home directory.
[REMOTE_CODE_EXECUTION]: The skill facilitates the use of an agentic coding tool with the capability to execute arbitrary shell commands and modify local codebases. Given the tool is downloaded from an unverified source, this poses a high risk of unauthorized command execution or system compromise.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted codebase data (e.g., source files, git histories) to drive automated terminal and editing actions.
Ingestion points: Multi-file code analysis, git commit message generation, and pull request reviews.
Boundary markers: Absent; there are no instructions provided to the agent to ignore potentially malicious content within the files it analyzes.
Capability inventory: File system writes, shell command execution, and git workflow manipulation.
Sanitization: None; the skill does not specify any validation or sanitization for the content processed by the tool.

claude-code-cli-terminal-assistant