The Agent Skills Directory

[COMMAND_EXECUTION]: The skill dynamically assembles and executes shell commands for project environment setup (e.g., npm install, cargo build, pip install) and verification (e.g., npm test, pytest) based on the presence of specific repository configuration files.
[EXTERNAL_DOWNLOADS]: The skill uses established package managers to download project dependencies from official and well-known registries such as npmjs.com, pypi.org, and crates.io. These operations are performed as part of the standard development lifecycle.
[PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection by processing and acting upon files in the repository without sanitization. If an agent uses this skill on a repository containing malicious configuration files, it may inadvertently execute harmful scripts during the setup or testing phases.
Ingestion points: The skill identifies and processes project files including package.json, Cargo.toml, requirements.txt, pyproject.toml, and go.mod to determine which commands to run.
Boundary markers: The skill lacks explicit instructions or delimiters to ensure the agent ignores potentially malicious commands embedded within the processed data files.
Capability inventory: The skill utilizes shell execution, file system access, and network capabilities (via package managers) to perform its tasks.
Sanitization: No verification or filtering of the content within the repository files is conducted before the commands are executed.

using-git-worktrees