The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides the agent with specific shell commands to create directories, manage Git repositories (clone, fetch, worktree), and interact with Maven. These commands are used to establish a persistent cache in the user's home directory at $HOME/.agents/librarian/.
[EXTERNAL_DOWNLOADS]: The skill is designed to download content from the internet using webfetch, git clone, and mvn dependency:get. It targets well-known services such as GitHub and Maven Central (e.g., fetching artifacts from org.typelevel and repositories from github.com).
[DYNAMIC_EXECUTION]: The agent is instructed to dynamically generate pom.xml files using shell redirection to facilitate dependency tree analysis and artifact retrieval via Maven.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from web pages, public Git repositories, and Maven metadata, which could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Data enters the agent's context through webfetch output, cloned source code files, and Maven dependency metadata (POM files, dependency trees).
Boundary markers: The skill does not explicitly define delimiters or instructions for the agent to ignore potentially embedded commands in the researched material.
Capability inventory: The skill has filesystem access to its cache directory and network access via multiple protocols (HTTP, Git, Maven).
Sanitization: There are no specified sanitization routines for the external content before it is processed or presented to the agent.

external-research