The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill orientation process involves reading and summarizing files from the current project, which could contain instructions intended to override agent behavior.
Ingestion points: The skill reads project manifests (e.g., package.json, composer.json), configuration files, READMEs, and representative source code files (documented in SKILL.md).
Boundary markers: The instructions do not specify the use of delimiters or clear warnings for the model to ignore instructions embedded within the files being read.
Capability inventory: The skill leverages Bash, Read, Write, Glob, and Grep tools to fulfill its functions.
Sanitization: Content from project files is read and processed without explicit sanitization or filtering of potential prompt injection strings.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute git-related commands for state management and change detection.
Evidence: The skill runs git rev-parse --git-common-dir to identify the cache location and git diff --name-status to perform delta updates on the project summary (SKILL.md).
[EXTERNAL_DOWNLOADS]: The documentation provides instructions to download and install the skill from an external repository.
Evidence: The README provides a command to clone the skill from the author's GitHub repository: git clone https://github.com/tommcfarlin/claude-code-onboard.git (README.md).

onboard