The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill creates a vulnerability surface by requiring the agent to fetch and prioritize instructions from an external repository defined in a local .project-brain.json file.
Ingestion points: The agent is instructed to read agent-guide.md, wiki pages, and source files from a remote GitHub repository into its context before performing substantive work.
Boundary markers: Absent. The instructions do not define delimiters or provide warnings to the agent to treat the ingested external content as untrusted or to ignore embedded instructions.
Capability inventory: The skill provides access to the GitHub CLI (gh) for repository cloning, API interaction, and creating pull requests, which could be abused if the agent follows malicious instructions from the "brain" repository.
Sanitization: Absent. Content from the external repository is processed directly to ground the agent's reasoning without validation or filtering.
[COMMAND_EXECUTION]: The skill utilizes shell commands via the GitHub CLI (gh) to check authentication status, clone repositories, and interact with the GitHub API.
[EXTERNAL_DOWNLOADS]: The skill fetches raw content and clones repositories from GitHub based on the repository field found in the .project-brain.json configuration file.

project-brain