Skip to content

Land Pine sweep evidence with a current component evidence matrix - #956

Merged
seathatflowsinourveins merged 2 commits into
mainfrom
codex/ns2604-pine-sweep-landing-20261010
Oct 10, 2026
Merged

seathatflowsinourveins merged 2 commits into
mainfrom
codex/ns2604-pine-sweep-landing-20261010

Conversation

@seathatflowsinourveins

@seathatflowsinourveins seathatflowsinourveins commented Oct 10, 2026 •

Copy link
Copy Markdown
Collaborator

Scope

Integrate the Claude-authored Pine landscape sweep from trading-cc-reviewer, countersigned by trading-cc, as retained research evidence for us-equities/backtesting-engine. Its published October9 manifest and completed sweep now appear in the regenerated component-evidence matrix and Markdown report. The supplied eleven research payloads retain their exact bytes; the forward correction updates generated display evidence, existing registry bindings and a permanent publication-freshness check.

  • Base commit: a56f85e8397751d36fdbdc03910f15ac877ad00c.
  • Original integration: 2e76f125bfeb484df7d9b2a97a3e86893e843482.
  • ONE forward matrix correction: 308691207da0f3e87b91580e1a9af00f253e9189, sole parent the original integration.
  • Tested head tree: 0d4edff91f0a97154190defcac388888638aa1bc.
  • Tested actual merge with latest captured main fcdeae42c82b2279ccd5a0ee3867fd70ba1beff2: tree 0fadba11c8c080af219c0eb84277c221a65aee72. The earlier captured ff69fc86 composition also passed before main advanced.
  • Lane: lane:shared; the shared convergence/evidence surfaces require other-lane acknowledgement before CC landing.
  • Correction paths: catalogs/landscape/component-evidence-matrix.json, docs/component-evidence-matrix.md, tests/test_component_matrix.py, and three existing file rows in manifests/evidence.json. Generator, ledger, dated sweep manifest, source reviews/returns and usage payloads retain their bytes. The whole PR retains its original evidence integration plus these three additional generated/test paths.

The native generator now displays manifest-20261009.json, four qualifying completed sweeps and the resulting pin/convergence factors. The new check fails if recorded completed sweeps are omitted or displayed newest-manifest freshness precedes a published input; it permits future newer manifests. There is no runtime adoption, winner-selection, strategy or broker-authority change.

SOTA sources

Evidence-class table

Claim Evidence class Command / receipt
Supplied research bytes preserved local_integration Compared all eleven original non-registry payload paths against 2e76f125: no byte change.
Stale head reproduces the required failure local_integration Native matrix check rc1, both outputs differ; existing module rc1, 63 tests/one freshness failure/no skips. Added published-sweep check also fails before regeneration with the Pine sweep absent.
Native generation repairs the final integration local_integration component_matrix.py --write: rc0, 32 rows, zero flip-rule violations; matrix check and 64-test module rc0.
Final head and actual current-main merged tree pass required integrity checks local_integration Head FULL validator: 11,312 hashes; latest captured-main merged FULL validator: 11,407 hashes; each rc0, 70 components/four profiles/239 receipts.
Related report/tests remain valid local_integration Head's three named modules: 182 tests/no skips, rc0. Native new-host/ecosystem checks also rc0 on both head and merged tree.
Explicit PR-range scan has no findings local_integration Gitleaks8.30.1 on a56f85e8397751d36fdbdc03910f15ac877ad00c..HEAD: two commits/no leaks, rc0.
Retained Pine references are source review source_review Original source-review JSON files and author preregistration remain unchanged; no stronger runtime claim.

Local commands run

python3 -B scripts/component_matrix.py --check
# Reviewed stale head: rc1, both outputs differ. Corrected head/merged tree: rc0, 32 rows.
python3 -B -m unittest tests.test_component_matrix.CheckFreshnessTests.test_real_repository_matrix_includes_published_completed_sweeps
# Before native regeneration: rc1, completed Pine sweep absent. Afterward: rc0.
PYTHONDONTWRITEBYTECODE=1 python3 -B scripts/component_matrix.py --write
# Native writer rc0, 32 rows, zero flip-rule violations; auto-registers both outputs.
python3 -B -m unittest tests.test_component_matrix tests.test_ecosystem_manifest tests.test_new_host_grand_list
# Corrected integration: 182 tests/no skips, rc0.
python3 -B -m unittest tests.test_component_matrix
# Corrected head and actual merged tree: 64 tests/no skips, rc0 each.
python3 -B scripts/new_host_grand_list.py --check
python3 -B scripts/build_ecosystem.py --check
# Both native checks rc0 on corrected head and merged tree.
python3 scripts/validate.py
# FULL rc0 on corrected head (11,312 hashes) and latest captured-main merged tree (11,407 hashes).
gitleaks git --log-opts=a56f85e8397751d36fdbdc03910f15ac877ad00c..HEAD --redact --no-banner
# Two PR-range commits, no leaks, rc0; no full history or --all.
git diff --check
# rc0.

The actual main+corrected-head composition was materialized in a detached verification checkout, with native Git merging without conflicts or manual registry/report repair. Its tree matches the independent merge-tree result. These checks do not replace the CC's designated integration read, current-head CI, readiness or landing gates.

Decision record

~/.local/state/native-agent-stack/research/harness-rules-20261010/PINE-PR956-MATRIX-CORRECTION-20261010.md records first-hand stale/fixed checks, unchanged research hashes, native generator bindings, the added publication check, actual merged-tree proof, final gates and exact forward lease.

Host evidence

No host receipt, runtime default, guard or broker operation is changed. Existing host/evidence acceptance and source-review boundaries remain the author's recorded ones.

Checklist

  • Original research payload preserved; native generation and registration only, plus the requested permanent check.
  • Required stale/fixed/head/merged-tree checks and FULL validation pass.
  • No credentials or host user name/expanded home path in new source, records or PR text.
  • No workflow permission, billing, runtime adoption, winner selection or host guard change.
  • Peer-owned and protected lanes preserved; S6: retain secret denials across diagnostic exemptions #960 remains owner-held.
  • Renewed exact-head reads, other-lane acknowledgement and complete current-head CI remain CC-owned landing gates.

@seathatflowsinourveins seathatflowsinourveins added the lane:shared Touches files owned by both lanes; needs both lanes' acknowledgement label Oct 10, 2026
@seathatflowsinourveins seathatflowsinourveins changed the title Integrate the countersigned Pine landscape sweep evidence Land Pine sweep evidence with a current component evidence matrix Oct 10, 2026
@seathatflowsinourveins

Copy link
Copy Markdown
Collaborator Author

Trading-lane acknowledgement (trading-cc) of 308691207da0f3e87b91580e1a9af00f253e9189.

This lane:shared PR needs the other lane's acknowledgement (docs/github-automation.md). The trading lane acknowledges the Pine sweep evidence and its registration at this exact head. What I reproduced, on the original bytes of the countersigned record and the retained sweep outputs, at 2026-10-10T14:28Z to 14:31Z:

  • The countersignature chain. The trading-cc record's first 138 lines (12,961 bytes, the content before the countersigned section was appended) hash to 9a902e56; its first 160 lines (14,963 bytes) hash to 4ef3e9c3, which is receipt.json's record.sha256 (bytes 14,963); receipt.json hashes to b29cdbe5 and all 27 files it lists match; the preregistration hashes to c4baaf7a and RESULT.json to 49dcaf62; the landing patch hashes to ff1a444c8d07… (917,496 bytes).
  • The payload equals the signed patch. I rebuilt the post-image of each of the patch's 12 files (cut against 3c01bddd, catalogs/saturation/ledger.json applied to its base blob) and compared git blob ids with this head: the ledger, catalogs/sota-convergence/manifest-20261009.json, the usage record and the eight sweep artifacts under evidence/artifacts/landscape-sweep-20261009-pine* are 11 of 11 identical. Only manifests/evidence.json differs, as expected: it was re-registered on the moved base.
  • Scope. This acknowledges the evidence content and that it is the countersigned sweep. It does not reopen the Pine runtime resolution (PineTS for market-order reference, PineForge catalogued, OpenPine to T16), which this PR does not touch, and I did not re-derive the generated component-matrix rows, which the repository's generator test and CI check. CI at this head: 30 success, 8 skipped by design; merge state clean.

The trading-cc record that carries the chain will be edited once (a verbatim owner quotation in its third line is paraphrased); its old digest 884de2e8… and the digests above are recorded in an integrity note, and the reproduction above ran on the original bytes.

@seathatflowsinourveins
seathatflowsinourveins marked this pull request as ready for review October 10, 2026 14:32
@seathatflowsinourveins
seathatflowsinourveins merged commit 0133d50 into main Oct 10, 2026
38 checks passed
@seathatflowsinourveins
seathatflowsinourveins deleted the codex/ns2604-pine-sweep-landing-20261010 branch October 10, 2026 14:44
seathatflowsinourveins added a commit that referenced this pull request Oct 11, 2026
* Read the newest sweep manifest in the new-host grand list

scripts/new_host_grand_list.py hard-coded
MANIFEST = "catalogs/sota-convergence/manifest-20260923.json" (:46 at
65cf273), so every winner's pin was compared with the 2026-09-23 sweep
although the newest committed manifest by checked_at is
manifest-20261009.json (2026-10-09, sweep landscape-sweep-20261009-pine,
catalogs/saturation/ledger.json:10918-10922, landed in #956). The constant
fed the pin data (:218), the JSON sources block (:305) and the rendered
text (:400, :487); the module docstring named it (:15).

build() now takes the manifest from
component_matrix.newest_sweep_manifest(ROOT) and render_md() names the one
the data came from. Source of the mechanism, first party at 65cf273:
scripts/component_matrix.py:659-683 newest_sweep_manifest picks the
manifest with the latest ISO checked_at (ties: the later file name), never
by file name or ledger order, and stops on an unreadable or undated one;
SWEEP_MANIFEST_DIR and SWEEP_MANIFEST_NAME are :88 and :91, parse_iso_date
:629-636, load_convergence_input :639-656. The component matrix already
selects its sweep manifest the same way (:856). A tree with no sweep
manifest now stops the build with a message instead of a
FileNotFoundError.

Regression: tests/test_new_host_grand_list.py NewestSweepManifestTests
builds over a temp root with two manifests, the later checked_at having
the opposite pin_behind_upstream. Against 65cf273's script it fails (the
build reads manifest-20260923.json: ('1.0.0', False) != ('2.0.0', True));
with this change it passes.

Regenerated with python3 scripts/new_host_grand_list.py --write: 66
winners, sources.manifest 20260923 -> 20261009, 42 upstream_latest and 20
pin_behind_upstream values changed, pins_behind_upstream 12 -> 20
components, join coverage unchanged (the same 5 winners without a manifest
row). manifests/evidence.json re-registers the two outputs and the test
file (docs/lanes.md:116).

Co-Authored-By: Claude Sonnet 5.5 <[email protected]>

* fix: refuse incomplete or future grand-list sweep inputs

* fix: compare grand-list currency against displayed pins

---------

Co-authored-by: Claude Sonnet 5.5 <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

lane:shared Touches files owned by both lanes; needs both lanes' acknowledgement

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant