The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands for diagnostic and repair purposes.
Evidence includes calls to npx nx for starting Metro and running tests, ls for checking build artifacts, and grep/cat for inspecting configuration files.
The skill also includes a file-writing operation using cat > ... << 'EOF' to create or update test setup files.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface where untrusted data could influence agent actions.
Ingestion points: User-provided error logs, Metro bundler output, and code snippets provided during the "Phase 1: Error Detection & Classification" step in SKILL.md.
Boundary markers: None identified. The skill does not explicitly instruct the agent to use delimiters or ignore instructions embedded within the error logs.
Capability inventory: The agent can execute shell commands (npx, npm, ls, grep, cat, git) and write to the filesystem (cat > ...) as seen in SKILL.md and resources/debugging-workflows.md.
Sanitization: There are no instructions for sanitizing or escaping variables derived from user input (e.g., {app-name} or [test name]) before they are interpolated into shell commands.
[EXTERNAL_DOWNLOADS]: The skill suggests installing development dependencies when missing.
Evidence in resources/debugging-workflows.md shows the command npm install --save-dev @babel/preset-react @babel/preset-typescript.
These packages are standard, well-known development tools from the official NPM registry and do not pose an inherent risk.

react-native-debug