The Agent Skills Directory

[COMMAND_EXECUTION]: The verification script scripts/verify-app.mjs executes shell commands using spawnSync to run tsc (TypeScript compiler), vite (build tool), and a structural validation script. These are standard operations for a development-focused skill.
[PROMPT_INJECTION]: The instructions contain directives that discourage the agent from inspecting the internal logic of the verifier or its delegated tools ('Do not read the verifier or its delegated validator'). This is a workflow constraint intended to enforce a specific pattern, rather than a malicious bypass of safety protocols.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits a vulnerability surface where it ingests user-controlled source files (e.g., App.tsx) and processes them via regex and build tools (tsc, vite).
Ingestion points: Source files are read in scripts/verify-app.mjs using readFileSync.
Boundary markers: No explicit boundary markers or 'ignore' instructions are used when processing file content.
Capability inventory: The script has the ability to execute subprocesses via spawnSync.
Sanitization: Content is validated via regex patterns but is not sanitized before being passed to build tools.

uit-contextual-inbox-app-web