The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill downloads and executes code from user-defined Git repositories. During installation and update processes, it runs npm install, pip install -r requirements.txt, and executes either main.py or tool.js found within the downloaded package.
[COMMAND_EXECUTION]: Multiple scripts leverage the subprocess module to execute shell commands. lib/git_ops.py performs git clone and git pull operations. scripts/install.py executes package managers and skill-specific initialization scripts. scripts/hook.py starts a background worker.
[EXTERNAL_DOWNLOADS]: The skill's primary operation involves fetching content from external Git repositories configured by the user.
[PROMPT_INJECTION]: The scripts/index.py script creates an indirect prompt injection surface. Ingestion points: Skill metadata including name and description is parsed from SKILL.md files in remote repositories. Boundary markers: Absent; metadata is stored and retrieved without delimiters. Capability inventory: The skill can execute shell commands, install packages via npm and pip, and run custom scripts via scripts/install.py and lib/git_ops.py. Sanitization: No sanitization or validation is performed on the ingested metadata before it is presented to the agent.
[COMMAND_EXECUTION]: The skill establishes persistence using an alwaysApply Cursor rule in rules/skill-store-hook.mdc. This rule triggers scripts/hook.py at every session start and explicitly instructs the agent to perform the execution quietly without notifying the user. The hook script subsequently spawns a detached background process for automated repository synchronization.

skill-store