docs(samples): add memory-aware-agent sample

[mjnovice]

Summary

Adds a new sample at packages/uipath/samples/memory-aware-agent/ that demonstrates how to wire UiPath Agent Memory into a coded agent.

The sample:

1. Searches a memory space via UiPath().memory.search_async
2. Stitches the LLMOps-rendered system_prompt_injection into the system prompt
3. Calls the LLM Gateway with the augmented context

Motivation: the Memory Spaces API landed recently (packages/uipath-platform/src/uipath/platform/memory/) but has no end-user sample. The SDK does not auto-wire memory based on the is_agent_memory_enabled flag in agent definitions — recall and prompt injection are explicit calls in agent code. The README spells that out so users do not look for magic that does not exist.

Files:

• main.py — async main(input: AgentInput) -> AgentOutput, traced
• pyproject.toml, uipath.json, bindings.json — match hello-world/llm_chat_agent conventions
• README.md — prereqs, run instructions, data-flow diagram

Test plan

• python -c "import ast; ast.parse(...)" — parses
• uvx ruff check . — passes (samples are excluded from lint, but the code is clean)
• uvx ruff format --check . — passes
• Resolved real imports in the package's uv env: uipath.platform.UiPath, uipath.platform.memory.{MemorySearchRequest, SearchField, SearchSettings, SearchMode}, uipath.platform.chat.ChatModels, uipath.tracing.traced — all present
• AgentInput / AgentOutput Pydantic models construct cleanly
• End-to-end run: requires UIPATH_URL, UIPATH_ACCESS_TOKEN, UIPATH_FOLDER_KEY, and a real memory space — left for a reviewer with credentials

🤖 Generated with Claude Code