The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The documentation references standard installation procedures for the official Temporal SDKs (temporalio for Python and @temporalio/* for TypeScript) and related AI integration packages from official registries (npm and PyPI).
[INDIRECT_PROMPT_INJECTION]: The skill describes patterns for AI agents (e.g., aiAgentWorkflow and ToolCallingAgent in ai-patterns.md) that ingest data from external sources such as user input, tool results, and API responses. While this is the intended function of an AI agent, it represents an inherent surface area for indirect prompt injection if the system processes untrusted data.
Ingestion points: External data enters the system through Workflow input parameters and Activity return values (referenced in ai-patterns.md and typescript-sdk.md).
Boundary markers: The provided code snippets demonstrate how to pass data between Workflows and Activities but do not include specific delimiters or instructions to ignore embedded commands within the LLM activities.
Capability inventory: The orchestration patterns described include capabilities for network operations (call_llm, fetch_data), tool execution, and potential file system access (write_file_activity).
Sanitization: The reference materials focus on durability and reliability; users are responsible for implementing input validation and sanitization when integrating LLMs with external capabilities.

temporal