The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill utilizes the !command syntax to execute shell commands at load time for environmental awareness.
The commands ls, test, and jq are used to identify directory structure, locate orientation documentation, and extract the project name from package.json.
These operations are confined to the local repository directory and do not involve sensitive system files or network communication.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and summarize content from untrusted repository files, which serves as a vector for indirect instructions.
Ingestion points: The skill reads README.md, AGENTS.md, CLAUDE.md, CONTRIBUTING.md, package.json, and neighboring source code files (SKILL.md).
Boundary markers: The instructions do not define delimiters or provide specific warnings to the agent to disregard instructions found within the repository files.
Capability inventory: The skill restricts itself to research and mapping, explicitly stating that no file edits should occur during the onboarding process.
Sanitization: There is no mechanism described to sanitize or filter the content of the orientation documents or source code before the agent processes them.

onboard-repo