The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and review potentially untrusted data from local files, specifically structured output schemas and JSON fixtures.
Ingestion points: The agent reads files from schemas/structured-output/** and src/snapshot-tests/__fixtures__/json/**.
Boundary markers: The instructions do not specify any delimiters or warnings to the agent to ignore instructions that might be embedded within these data files.
Capability inventory: The agent has the capability to execute shell commands (npm, npx) and read local files.
Sanitization: No sanitization or pre-processing steps are mentioned for the content being reviewed.
[COMMAND_EXECUTION]: The skill instructs the agent to run several local shell commands for validation purposes.
Evidence: Commands include npm run test:schema-fixtures, npm test -- src/core/__tests__/structured-output-schema.test.ts, npm run typecheck, and npx skill-check .agents/skills/mobilebuildmcp-structured-output-review.
These are standard development workflows, but they provide a mechanism for local code execution if the environment or package configurations are compromised.
[EXTERNAL_DOWNLOADS]: The skill directs the agent to inspect content from an external source, specifically documentation pages hosted on xcodebuildmcp.com.
Evidence: xcodebuildmcp.com/app/docs/_content/schema-versioning.mdx and xcodebuildmcp.com/app/docs/_content/output-formats.mdx are listed under 'Files to inspect'.
Fetching and processing content from external domains can introduce instructions or data that influence agent behavior.

mobilebuildmcp-structured-output-review