Repository navigation
Retain the SPY/LEAN parity verdict on main (gate G-a stays BLOCKED) - #67
Merged
seathatflowsinourveins merged 7 commits intoSep 22, 2026
Merged
Conversation
… rows Land the reviewed SPY/LEAN parity harness (Gate G-a) under blueprints/us-equities/engine-nautilus/spy-parity/ with tests/test_spy_parity.py, so the gate ladder's receipt_path resolves on main. The verdict stays BLOCKED (complete, 29 checks, 4 failed, all attributed to the two mappings the bound manifest declares unsupported: market-on-open and dividend cash posting). Host paths were removed from the manifest, one probe and the README; because the manifest is hash-bound, the receipt was regenerated by the documented bwrap two-run replay (fills, dividend cash and end cash unchanged) and compare.py re-derived the verdict. Only the two gate rows owned by this lane change (notes); statuses are unchanged and the checker reports no flip candidates. Checks: tests.test_spy_parity OK; unittest discover 1464 OK (204 skipped); scripts/validate.py passed; validate_catalogs.py ok; trading_gates.py --check passed; guarded gitleaks on the staged change: no leaks. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
Review of PR #67 found four issues in this lane's additions: - the dividend-sim-module note claimed a probe records a SimulationModule cash-posting path and suggested a synthetic open tick; the probes record neither, and the bound manifest calls synthetic open data fabrication. The note now states only what the probes record. - the spy-lean-parity note referred to review findings in the README that were not there; the README now lists the six carried findings. - "byte-identical" re-derivation is now recorded with the verdict sha256. - none of the landed parity files were hash-pinned in manifests/evidence.json; 16 rows now pin the harness, probes, receipt, verdict and tests. Checks: tests.test_spy_parity OK; python3 -m unittest 1464 OK (204 skipped); every hosted validate step and trading_gates.py --check pass; guarded gitleaks: no leaks. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
Re-review of PR #67 found the dividend-sim-module gate and the harness README still advertised a dividend SimulationModule and a synthetic open tick, both of which the bound mapping manifest refuses as fabricating the mapping, and that the "no ingestible dividend type" finding was cited to the raw probe rather than the manifest's conclusion. The gate title, note and README now state that closing the gate needs an engine-native mechanism or a newly preregistered manifest. A repin helper had also rewritten adaptive-paper/source-hashes.json, outside this PR's scope; it is restored. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
…verdict Accuracy check of 458c049 found the dividend-sim-module note still opened with a pre-existing "Designed" claim that contradicted its own conclusion, and credited the cash-entry refusal to the wrong source. The note now opens with "No manifest-accepted design exists yet" and credits the cash refusal to the manifest's decision and acceptance-plan.md, the open-data refusal to the manifest. The README credits the review record in the agent-lab task record. The engines-strategies card (JSON and Markdown) still named a dividend SimulationModule and said the per-check total was not retained; it now points to the retained verdict file and the gate. PR-4's catalog test that pinned the old "parity worktree" wording now asserts the retained verdict path instead. Checks: every hosted validate step, trading_gates.py --check and python3 -m unittest pass; guarded gitleaks: no leaks. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
Resolves the explorer conflict by regenerating docs/ecosystem/index.html and repinning its evidence row; all hosted validate steps and 1679 unit tests pass. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
…-evidence-20260922
The fix-verification review found the README's 21-of-24 count cited a task record that is not in this repository; it is now labelled external, unpinned evidence, and the six items remain restated against named files. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
seathatflowsinourveins
added a commit
that referenced
this pull request
Sep 22, 2026
Explorer regenerated and the gate-file and explorer evidence rows repinned; all hosted validate steps, trading_gates and 1987 unit tests pass. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
seathatflowsinourveins
deleted the
claude/trading-parity-evidence-20260922
branch
September 25, 2026 18:47
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
catalogs/us-equities/gates-20260922.json→spy-lean-parity.receipt_pathresolves on main. Verdict: BLOCKED, complete, 29 checks, 4 failed, all attributed to the two manifest-declared unsupported mappings (market-on-open fill; dividend cash posting). 0 unattributed / skipped / rejected.spy-lean-parity,dividend-sim-module) — notes only, statuses unchanged; the checker reports no flip candidates.bwraptwo-run replay (fills, 428.64 dividend cash and 90078.96 native end cash unchanged) andcompare.pyre-derived the verdict.Evidence class
HIST (deterministic local replay against the frozen LEAN 985ef30 oracle). Not upstream E2E, not paper.
Checks run
python3 -m unittest tests.test_spy_parity→ OKpython3 -m unittest discover -s tests -q→ 1464 OK (204 skipped)python3 scripts/validate.py→ passed;python3 scripts/validate_catalogs.py→ okpython3 scripts/trading_gates.py --check→ passed, no flip candidatesgitleakson the staged change → no leaksReview
Prior independent review of the harness: 21/24 claims confirmed, six non-blocking hardening items (commission read-back fallback, in-window-only value validation, conditional iteration check, dropped nonpositive distribution, README mode claims, inventory count). A fresh review of this PR's diff follows before merge.
🤖 Generated with Claude Code