Skip to content

Phase 3 — structured L3 review panels #9

Description

@thebeastagi

Scope (ROADMAP.md, Phase 3)

  • Role-specialized reviewer schemas: statistician, domain theorist, methodologist, devil's advocate.
  • Multi-round reports, objection tracking, disagreement gates.
  • Catch-rate scoring on seeded known-bad and known-good candidates.

Acceptance checklist

  • Structured reviewer schemas for the four roles, each emitting typed verdicts (approve / object / abstain) with severity-graded objections and rationales.
  • Multi-round panel protocol: independent round 1, rebuttal rounds on disagreement, objection lifecycle tracking (raised / withdrawn / sustained).
  • Disagreement gates: blocking objections prevent L3 promotion until resolved; deterministic reference reviewers for the synthetic domains.
  • Panel verdict is recorded as review evidence through the kernel/ledger (no side channels).
  • Catch-rate scoring harness over Phase-2 seeded fixtures: report known-bad catch rate and known-good false-reject rate.
  • Tests green (ruff + pytest 3.10/3.11/3.12). Docs updated truthfully.
  • Invariants: L4 human-gated; no discovery claims; no secrets; $0 spend.

Dependencies

Blocked by #8 — Phase 2 epic (panels consume validation gates, seeded fixtures, and calibration reports).

Tasks

Metadata

Metadata

Assignees

No one assigned

    Labels

    epicPhase epic: groups task issuesphaseRoadmap phase marker

    Projects

    No projects

    Milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions