FPRD ChatGPT research rollup

Prepared: 2026-07-29
Purpose: handoff to a repository-owning Codex run for comparison, audit, and coherent integration into insight-forge/boxlab/fprd-lab.

This archive collects every FPRD research artifact from the recent ChatGPT runs that could be recovered from the current workspace or durable file storage. It includes reports, theorem drafts, hostile audits, literature comparisons, computational checkers, handoff notes, and the earlier causal/origami source bundle in both expanded and original form.

Contents
  1. RECOVERY-STATUS.md
  2. RUN-SUMMARIES.md
  3. 00-handoff-and-prior-guidance/FPRD-Lab-clean-chat-handoff-2026-07-29.md
  4. 01-causal-origami-source/INDEX.md
  5. The numbered work-cycle directories in order
  6. MANIFEST.sha256

Directory map

  • 00-handoff-and-prior-guidance/: prior handoffs and audit packets.
  • 01-causal-origami-source/: expanded causal factorization, sewing, and Fold-and-Look source bundle.
  • 02-horizon-join-and-origami-image/: strong negative lifting result, common-horizon frontier, and one-hinge image work.
  • 03-adaptive-access-and-reservations/: access entropy, causal reservation transfer, diagonal obstruction, and associated finite searches.
  • 04-wfl-residual-row/: exact cycle summary and recovery status for the WFL residual-row capacity run.
  • 05-derived-control-and-cocycle/: derived prefix-control construction, proof, audit, and exhaustive checkers.
  • 06-broader-fprd-probes/: the two cross-area probe sets, arithmetic shape probe, and hostile framework audit.
  • 07-site-provenance/: deployed-site version and source-commit record.
  • original-bundles/: byte-for-byte copy of the earlier source ZIP.

Integration warning

These files are research inputs, not an instruction to overwrite the repository. The repository’s current proof and audit artifacts remain authoritative. Compare each result against current CLAIMS.yaml, docs/EVIDENCE.md, PLAN.md, OPEN_PROBLEMS.md, proof artifacts, audits, and the actual current branch.

Several claims were independently audited or exhaustively checked, but none should be promoted solely because it appears in this bundle. Preserve the scope qualifications in each packet, especially:

  • the lifting counterexample depends on one stationary presentation sharing a layout across horizons;
  • WFL global Look is charged through old-dependent fiber/site incidence, not as a bounded local read;
  • the derived-control result is for a pinned separated old/query-site doctrine;
  • the rooted-archive height calibration is not yet the repository’s tropical canonical-height theorem;
  • literature searches delimit comparator status but do not establish novelty.

Verification

VERIFICATION.md records which Python artifacts were executed while assembling this rollup. MANIFEST.sha256 gives a checksum for every file other than the manifest itself.