fature: Multi context smart agent + multi context validation

[carlesonielfa]

Summary

• Preserve one Context per Smart Agent tool call instead of flattening results into one synthetic context.
• Add hierarchical multi-context validation with per-context answers and citations plus a global missing-information query.
• Prune each context independently while retaining summary-only contexts and dropping irrelevant contexts.

Tests

• Add deterministic and cassette-backed coverage for relevance, summary-only answers, chunk pruning, and missing information.

[Copilot]

Pull request overview

This PR introduces end-to-end multi-context handling: SmartAgent now preserves each tool-returned Context independently, and ContextAgents can validate/prune multiple contexts in a single pass with per-context partial answers/citations plus a global “missing information” query.

Changes:

• Add multi-context validation prompt + JSON schema and a validate_contexts_and_answer / save_contexts_and_return_missing flow to prune/save contexts independently.
• Update SmartAgent to collect and return a list of contexts (instead of flattening into one synthetic context), and adjust tests accordingly.
• Add deterministic + VCR cassette-backed tests covering relevance filtering, summary-only contexts, chunk pruning, and global missing-info reporting.

Reviewed changes

Copilot reviewed 12 out of 12 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
hyperforge/src/hyperforge/context/agent.py	Adds multi-context validation + saving/pruning logic.
hyperforge/src/hyperforge/prompts.py	Introduces the multi-context validation prompt template + JSON schema.
hyperforge/src/hyperforge/memory/memory.py	Extends save_context to optionally index contexts under a parent agent id.
agents/smart/src/hyperforge_smart/agent.py	Switches SmartAgent to return and retain multiple contexts per tool call.
agents/smart/tests/test_smart_mcp_perplexity.py	Updates expectations to handle multiple returned contexts.
agents/smart/tests/test_contexts.py	Adds unit test ensuring tool results preserve original contexts.
agents/summarize/src/hyperforge_summarize/agent.py	Adds a TODO note about exposing summaries when forcing chunk citations.
hyperforge/tests/context/test_validation.py	Adds multi-context validation tests (mocked + VCR-backed).
hyperforge/tests/context/cassettes/test_validation/test_multi_context_validation_keeps_relevant_contexts.yaml	VCR cassette for relevance filtering behavior.
hyperforge/tests/context/cassettes/test_validation/test_multi_context_validation_keeps_summary_and_discards_chunks.yaml	VCR cassette for summary-only context retention behavior.
hyperforge/tests/context/cassettes/test_validation/test_multi_context_validation_prunes_irrelevant_chunks.yaml	VCR cassette for chunk-level pruning behavior.
hyperforge/tests/context/cassettes/test_validation/test_multi_context_validation_reports_globally_missing_information.yaml	VCR cassette for global missing-info query behavior.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.