The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill operates on external data from the project codebase and API specifications, which represents an attack surface for indirect instructions.
Ingestion points: The agent reads mobile source code (.swift, .dart, .ts, .tsx), configuration files (Package.swift, pubspec.yaml, package.json), and API specifications (openapi.yaml).
Boundary markers: No specific delimiters or boundary warnings are provided to separate project data from agent instructions.
Capability inventory: The agent has permissions to write to the codebase, modify platform configurations, and execute build and test tools.
Sanitization: The instructions do not include explicit sanitization or validation for data ingested from the project environment.
[COMMAND_EXECUTION]: The skill uses standard platform build and verification tools as part of its development workflow.
Evidence: The stack configurations define commands such as swift build, flutter test, dart analyze, and npx jest to verify code changes.
Context: These are legitimate and expected actions for a developer-oriented AI agent.
[EXTERNAL_DOWNLOADS]: The skill references official and trusted external resources for its technical stack.
Evidence: The Swift iOS snippets reference official Apple repositories for OpenAPI generation and runtime support.
Context: These are verified, trusted sources commonly used in mobile development.

oma-mobile