Skip to content

Phase 4 — real domain adapters (Kepler astronomy) #10

Closed

Description

@thebeastagi

Scope (ROADMAP.md, Phase 4)

  • ASTRA/GEODISC/BIODISC/SLATE adapters only after licence and owner review.
  • Domain-specific validators remain sandboxed behind adapters.
  • Cross-domain method transfer enters target domain at L0 every time.

Owner directive (Robin, 2026-07-19): build ONE real astronomy adapter using the existing Kepler-10 b path (public MAST light-curve data, already used by demos/ledger-grok). Clean-room: NO ASTRA/GEODISC/BIODISC/SLATE code — their licensing is unresolved and we hold no owner permission for code reuse.

Acceptance checklist

  • One real adapter (Kepler photometry) registered at the first-party real-data trust tier; ships/bundles the public NASA/MAST Q1 light curve with a provenance note (data, not code).
  • Adapter re-derives the known Kepler-10 b signal (bounded stdlib BLS path, ported from this repo's own Apache-2.0 demo code) — framed everywhere as validation of a published result, never a discovery.
  • Staged validators behind the adapter: internal (detection SNR), replication (independent holdout halves re-detect), review (adversarial: odd/even depth, secondary-eclipse, harmonic checks) — all sandboxed inside the adapter; the kernel sees only contract evidence.
  • Negative controls: no-signal and tampered curves fail the gates.
  • Cross-domain transfer of a Kepler candidate enters the target domain at L0 (bridge invariant re-verified against a real adapter).
  • Tests green (ruff + pytest 3.10/3.11/3.12), runtime-bounded for CI. Docs updated truthfully (incl. honesty framing).
  • Invariants: L4 human-gated; no discovery claims; no secrets; $0 spend; no ASTRA-family code.

Dependencies

Blocked by #7 — Phase 1 epic (trust registry + permission manifest) and #8 — Phase 2 epic (validation gates the adapter's evidence passes through).

Tasks

Activity

  1. added this to the Phases 1–4 milestone on Jul 19, 2026
  2. added
    phaseRoadmap phase marker
    epicPhase epic: groups task issues
    on Jul 19, 2026
  3. thebeastagi commented on Jul 20, 2026

    @thebeastagi
    OwnerAuthor

    All tasks complete and merged in #38. Clean-room Kepler adapter (CORE tier) re-derives published Kepler-10 b; negative controls; L0→L3 capstone; L4 refusal verified — v0.5.0.

    • All task issues closed (see checklist above)
    • CI green on merge (3.10/3.11/3.12)
    • Invariants verified: L4 human-gated, no discovery claims, clean-room, no secrets
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

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