feat: support conversational runtime prompts

[radu-mocanu]

Summary

• resolve conversational advanced system prompts from each exchange input
• preserve custom inputs, user settings, history, aliases, and internal field collisions
• preserve assistant attachments and existing static prompt behavior

Why

PR #1016 added runtime prompt interpolation for autonomous advanced agents. Conversational advanced agents need the same behavior so custom inputs and user settings are resolved for every turn.

Validation

• full pytest suite
• mypy across the repository
• Ruff lint and format
• two independent adversarial reviews

[Copilot]

Pull request overview

Adds runtime system-prompt interpolation to conversational advanced agents, aligning their behavior with autonomous advanced agents so per-turn exchange inputs (custom inputs/user settings) can be resolved into a system prompt once per invocation and applied to each Deep Agents model request, while preserving backward compatibility for static prompts.

Changes:

• Extend create_conversational_advanced_agent_graph to accept a callable system_prompt, resolve it from exchange inputs, and inject it via _RuntimeSystemPromptMiddleware with collision-safe state keys.
• Introduce a dedicated conversational input base model to preserve alias handling and message contract behavior, including support for user-supplied input_schema without dropping conversation messages.
• Add comprehensive tests covering callable vs static prompts, alias serialization, schema/message preservation, internal-field collision avoidance, and Deep Agents prompt composition.

Reviewed changes

Copilot reviewed 4 out of 5 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
src/uipath_langchain/agent/advanced/agent.py	Implements per-exchange runtime system prompt resolution for conversational graphs and injects it into Deep Agents requests via middleware with collision-safe state keys.
src/uipath_langchain/agent/advanced/types.py	Adds a conversational input base model with alias/name validation config to support reliable prompt input serialization and wrapper state composition.
tests/agent/advanced/test_conversational_advanced_agent_graph.py	Adds test coverage for runtime prompt resolution, alias handling, collision safety, and Deep Agents request augmentation.
pyproject.toml	Bumps package version to 0.16.2.
uv.lock	Updates lockfile to reflect version bump.

[sonarqubecloud]

Quality Gate passed

Issues
6 New issues
0 Accepted issues

Measures
0 Security Hotspots
99.2% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud