The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to gather information about the local environment and provides templates for CLI-based workflows.
It instructs the agent to run git remote -v and read .git/config to identify the repository host.
It establishes conventions for using the gh (GitHub) and glab (GitLab) CLI tools to interact with remote repository services.
[PROMPT_INJECTION]: The skill creates an attack surface for both direct and indirect prompt injection through its repository discovery and configuration process.
Meta-Instruction Injection: The skill modifies CLAUDE.md or AGENTS.md to include a new ## Agent skills block, which dictates the agent's future behavior and conventions for the project.
Indirect Injection Surface: The agent is instructed to read various repository files (AGENTS.md, CLAUDE.md, CONTEXT.md, package.json, etc.) to determine the project state.
Ingestion points: Data is read from multiple files at the repository root and within docs/ and src/ directories.
Boundary markers: No delimiters or instructions are used to separate untrusted repository content from the agent's control logic.
Capability inventory: The skill has permissions to read/write files and execute powerful CLI tools like git, gh, and glab.
Sanitization: There is no evidence of sanitization or validation of the ingested data before it is used to influence the setup logic.

setup-matt-pocock-skills