AutoResearch/Idea discovery subproject
Version library

VERSION / V4.2

Hill

Expose missing evidence at runtime

Historical protocol & software prototype

Hill focuses on engineering reliability. A model-call bridge and structural and semantic validation turn missing responses, placeholders and empty refutation ledgers into pauses or explicit invalid states, with review records saved separately.

Philosophy

An existing artifact does not mean the research is complete, and structural validity does not make content true. The execution system should expose unresolved work.

Architecture & control flow

  1. 01

    The bridge supports offline request/response fulfillment, CLI model invocation and an external-orchestrator interface.

  2. 02

    Validate nodes, refutation ledgers and reports structurally and semantically before writing.

  3. 03

    Pause unresolved calls; missing required ledgers or invalid content lead to an invalid report.

  4. 04

    Save review artifacts, run state and rejection information for diagnosis.

Architecture outline derived from this version’s control flow.

What this version changes

Adds runtime gates, SubagentBridge and SemanticValidator to the V4-pro scaffold, moving the focus from protocol completeness to exposing execution defects.

Implementation & evidence scope

Real content still depends on model responses. In this snapshot the claude_code branch needs external dispatch; node generation, live literature retrieval and complete end-to-end capability remain unresolved.

Code & bundled material

The introduction draws on bundled notes, changelogs and central code. Software tests, synthetic diagnostics and scientific effectiveness use different evidence standards.

Source references
  • novelty-idea-generator/V4_HILL_CHANGELOG.md · 19–51
  • novelty-idea-generator/harness/search/subagent_bridge.py · 66–103
  • novelty-idea-generator/harness/search/validators.py · 116–150
  • novelty-idea-generator/V4_HILL_CHANGELOG.md · 188–193