fix(review): publish substantive OpenCode LLM evidence

AGENTS.md

@@ -7,4 +7,7 @@ Materialize accepts only exact SHA-256 pins or a bounded relative `-r` include (
 Conflict-scope roots fail closed when the immediate parent directory is a symbolic link.
 OriginWeave hourly NVIDIA NIM repair is a thin caller at minute 10. See [`docs/doctoring/originweave-hourly-review-caller.md`](docs/doctoring/originweave-hourly-review-caller.md).
 nonnest2 hourly NVIDIA NIM repair is a thin caller at minute 16. See [`docs/doctoring/nonnest2-hourly-review-caller.md`](docs/doctoring/nonnest2-hourly-review-caller.md).
+
+OpenCode may repair only trusted `path:line` bindings on LLM probes that already carry an independent proof and source-line digest. See [`docs/doctoring/opencode-llm-review-publication.md`](docs/doctoring/opencode-llm-review-publication.md).
+
 The materialization contract is also covered by [`docs/doctoring/exact-artifact-sbom-attestation.md`](docs/doctoring/exact-artifact-sbom-attestation.md).

ARCHITECTURE.md

@@ -112,6 +112,9 @@ sequenceDiagram
 - Required review workflows execute **base-branch** scripts. A PR that edits
   those workflows cannot widen its own `pull_request_target` token.
 - Reviewer agents stay `edit: deny`. They judge; they do not implement.
+- OpenCode remains the review reasoner. Deterministic code may repair only
+  trusted `path:line` source-line digest bindings on LLM probes; it never
+  invents a hypothesis, observed result, or verdict.
 - Sandbox helpers copy the workspace, drop secret environment values unless
   explicitly allowlisted by **name**, and run subprocesses with `shell=False`.
 - Logs and review receipts redact credential shapes (tokens, bearer values,
@@ -145,6 +148,8 @@ trusted `uv` exporter is downloaded from the literal GitHub Releases URL for
   contract.
 - [`docs/doctoring/hourly-nvidia-nim-autofix.md`](docs/doctoring/hourly-nvidia-nim-autofix.md)
   — current increment's repair-worker decision and APA 7th citations.
+- [`docs/doctoring/opencode-llm-review-publication.md`](docs/doctoring/opencode-llm-review-publication.md)
+  — LLM probe publication without inventing observed proof.
 - [`docs/doctoring/fast-mlsirm-hourly-review-caller.md`](docs/doctoring/fast-mlsirm-hourly-review-caller.md)
   — product-specific psychometric repair heartbeat and scientific gates.
 - [`docs/doctoring/exact-artifact-sbom-attestation.md`](docs/doctoring/exact-artifact-sbom-attestation.md)

CHANGELOG.md

@@ -41,6 +41,7 @@ Semantic Versioning where the repository publishes a release.
 - Download the pinned `uv` 0.12.1 exporter from the official GitHub Releases URL instead of `releases.astral.sh`, which now returns HTTP 403 and blocks org-wide OpenCode `coverage-evidence`. The SHA-256 pin is unchanged. The opener may follow one hop onto `release-assets.githubusercontent.com` or `objects.githubusercontent.com` and still rejects every other host, userinfo, non-HTTPS scheme, and nondefault port (ContextualWisdomLab/.github#1109).
 - Compared the trusted `uv` executable's post-install `--version` output against the real GitHub Releases build's full string, `uv 0.12.1 (x86_64-unknown-linux-gnu)`, instead of the bare `uv 0.12.1` the prior check required; the genuine release binary always prints the target triple, so every installation was failing the pin check immediately after the archive download itself was fixed (ContextualWisdomLab/.github#1109).
 - Excluded relative `-r` and `--requirement` referrers from generated flat base-lock publication while retaining bounded include syntax diagnostics and discovering independently complete direct `.txt` children of `requirements` directories.
+- Published substantive OpenCode LLM probes when they already carried an independent proof and exact source-line digest but omitted a duplicated `path:line` citation, so NVIDIA NIM / OpenCode review evidence is no longer discarded as `NO_CONCLUSION`.
 - Refused a conflict-scope repository root whose immediate parent is a symbolic link, so a swapped parent cannot redirect the canonical worktree after the last-component check (CWE-367).
 - Materialized base Python locks only when every package line is an exact SHA-256 pin or a bounded relative `-r`/`--requirement` include. A lone `--require-hashes` directive, a dotted include such as `./lock.txt`, or `-r other-hashes.txt` no longer enters the trusted build context.
 - Bounded the Strix quality self-test's deterministic timeout fixtures to 3-second process and 5-second fake-sleep budgets so exact-head policy evidence completes inside the existing job limit without changing production Strix scanner timeouts, providers, credentials, or review semantics.

CLAUDE.md

@@ -37,7 +37,10 @@ This is the ContextualWisdomLab **organization-wide `.github` special repository
 **OpenCode judges PRs; GitHub Actions performs mechanical updates and merges.** OpenCode approval is
 evidence-gated (changed files, CodeGraph evidence, Change Flow DAG, test/coverage/docstring evidence,
 an actually-executed PoC via `scripts/ci/sandboxed_verify.py` or `scripts/ci/sandboxed_web_e2e.py`,
-split `Developer experience:` / `User experience:` sections). The scheduler updates a PR branch only
+split `Developer experience:` / `User experience:` sections). Deterministic
+code may repair only trusted `path:line` bindings on LLM probes that already
+carry an independent proof and source-line digest; it never invents observed
+results. The scheduler updates a PR branch only
 when the latest review is approved, no current-head check has failed, and GitHub reports the PR as
 behind. The mechanical merge scheduler itself never synthesizes a fix: it gives `DIRTY`/`CONFLICTING`
 PRs repair guidance. A separate edit-capable autofix flow

docs/doctoring/opencode-llm-review-publication.md

@@ -0,0 +1,45 @@
+# OpenCode LLM review publication
+
+## Decision
+
+OpenCode remains the review reasoner. Deterministic code may repair only the
+serialization binding between an LLM-authored adversarial probe's structured
+`path`/`line` and the SHA-256 of that exact immutable current-head source line.
+It must not create or alter the hypothesis, counterexample, observed result,
+outcome, finding, or verdict.
+
+## Root cause
+
+Production run `31310130263` invoked NVIDIA NIM and OpenCode free models, but
+several substantive control blocks were discarded even when they carried an exact
+source-line receipt because the prose omitted the duplicated `path:line` citation.
+Noema is intentionally dispatched only after an exact-head OpenCode approval,
+so this publication failure suppressed both review identities.
+
+## Safety boundary
+
+Repair is attempted only when original model evidence already contains an
+independent proof class and an observed result. The trusted current-head tree
+must resolve the structured path and line. A path whose POSIX parts include
+`..` is left unchanged (CWE-22; MITRE, n.d.). Changed-file, source-tree,
+runtime-receipt, duplicate-probe, outcome, finding-location, coverage,
+language, and publication gates still run afterward. Unsupported prose
+continues to return `NO_CONCLUSION`.
+
+## Verification
+
+The regression suite proves canonical path/line repair only when an exact receipt
+already exists, refusal to invent an observed result, rejection of missing or mismatched
+receipts, and preservation of malformed and
+unverifiable shapes, full normalizer behavior, branch coverage, docstrings,
+compileability, and a clean worktree.
+
+## References
+
+MITRE. (n.d.). *CWE-22: Improper limitation of a pathname to a restricted
+directory ('path traversal')*. Retrieved August 13, 2026, from
+https://cwe.mitre.org/data/definitions/22.html
+
+National Institute of Standards and Technology. (2020). *Security and privacy
+controls for information systems and organizations* (NIST Special Publication
+800-53 Rev. 5). https://doi.org/10.6028/NIST.SP.800-53r5

requirements-pip-audit-ci-hashes.txt

@@ -213,9 +213,9 @@ packaging==26.2 \
     # via
     #   pip-audit
     #   pip-requirements-parser
-pip==26.1.2 \
-    --hash=sha256:382ff9f685ee3bc25864f820aa50505825f10f5458ffff07e30a6d96e5715cab \
-    --hash=sha256:f49cd134c61cf2fd75e0ce2676db03e4054504a5a4986d00f8299ae632dc4605
+pip==26.2.1 \
+    --hash=sha256:71138adf1f4ca900cdb7d289c21b7494329f2332b6d85f0e1c42108c0384ed3e \
+    --hash=sha256:f6ad667e89a1fe78046c8f13232b247200f5258d7828f3f7883d660878e0813f
     # via pip-api
 pip-api==0.0.34 \
     --hash=sha256:8b2d7d7c37f2447373aa2cf8b1f60a2f2b27a84e1e9e0294a3f6ef10eb3ba6bb \

scripts/ci/opencode_review_normalize_output.py

@@ -653,6 +653,82 @@ def adversarial_probe_source_receipt_error(
     return ""
 
 
+def repair_adversarial_probe_source_bindings(value: dict[str, Any]) -> dict[str, Any]:
+    """Canonicalize only the trusted path and line citation of LLM probes.
+
+    The model remains solely responsible for the hypothesis, counterexample,
+    observed proof, outcome, finding, and verdict. Repair runs only when the
+    original model evidence already names an independent proof class, an
+    observed result, and the exact valid source-line digest from the immutable
+    current-head tree. Missing or mismatched digests remain rejected.
+    """
+    validation = value.get("adversarial_validation")
+    if not isinstance(validation, dict):
+        return value
+    probes = validation.get("probes")
+    if not isinstance(probes, list):
+        return value
+
+    repaired_probes: list[Any] = []
+    changed = False
+    for probe in probes:
+        if not isinstance(probe, dict):
+            repaired_probes.append(probe)
+            continue
+        path_value = probe.get("path")
+        line_value = probe.get("line")
+        evidence_value = probe.get("evidence")
+        if (
+            not isinstance(path_value, str)
+            or not path_value.strip()
+            or isinstance(line_value, bool)
+            or not isinstance(line_value, int)
+            or line_value <= 0
+            or not isinstance(evidence_value, str)
+            or not evidence_value.strip()
+        ):
+            repaired_probes.append(probe)
+            continue
+
+        normalized_path = path_value.strip()
+        if ".." in PurePosixPath(normalized_path).parts:
+            repaired_probes.append(probe)
+            continue
+        receipt_error = adversarial_probe_source_receipt_error(
+            evidence_value,
+            normalized_path,
+            line_value,
+        )
+        if receipt_error:
+            repaired_probes.append(probe)
+            continue
+        digest = SOURCE_LINE_RECEIPT_RE.findall(evidence_value)[0].casefold()
+
+        lexical_evidence = SOURCE_LINE_RECEIPT_RE.sub("", evidence_value).strip()
+        receipt_bound_evidence = (
+            f"{lexical_evidence} source-line-sha256={digest}"
+        ).strip()
+        if adversarial_evidence_rejection_reason(receipt_bound_evidence, ""):
+            repaired_probes.append(probe)
+            continue
+
+        canonical_evidence = (
+            f"{lexical_evidence} Trusted current-head source binding at "
+            f"{normalized_path}:{line_value}; source-line-sha256={digest}"
+        ).strip()
+        repaired_probes.append(
+            {**probe, "path": normalized_path, "evidence": canonical_evidence}
+        )
+        changed = True
+
+    if not changed:
+        return value
+    return {
+        **value,
+        "adversarial_validation": {**validation, "probes": repaired_probes},
+    }
+
+
 def adversarial_validation_error(
     value: Any,
     *,
@@ -1267,6 +1343,7 @@ def reject(reason: str) -> None:
         return reject("APPROVE cannot contain findings")
     if result == "REQUEST_CHANGES" and not findings:
         return reject("REQUEST_CHANGES requires at least one finding")
+    value = repair_adversarial_probe_source_bindings(value)
     adversarial_error = adversarial_validation_error(
         value.get("adversarial_validation"),
         result=result,

scripts/ci/test_strix_quick_gate.sh

@@ -1506,8 +1506,8 @@ assert_pr_review_merge_scheduler_uses_github_actions_bot_token() {
 	assert_file_contains "$workflow_file" "github.event_name == 'pull_request_target' && format('pr-{0}', github.event.pull_request.number)" "scheduler scopes pull_request_target concurrency to the active PR"
 	assert_file_contains "$workflow_file" "github.event_name == 'workflow_run' && github.event.workflow_run.pull_requests[0].number && format('pr-{0}', github.event.workflow_run.pull_requests[0].number)" "scheduler scopes workflow_run concurrency to the completed review PR"
 	assert_file_contains "$workflow_file" "github.event_name == 'schedule' && format('schedule-{0}', github.event.schedule)" "scheduler isolates the 15-minute organization sweep from the separate 30-minute scheduled scan"
-	assert_file_contains "$workflow_file" "github.event_name == 'repository_dispatch' && github.run_id" "scheduler keeps manual queue scans isolated per run"
-	assert_file_contains "$workflow_file" "cancel-in-progress: \${{ github.event_name == 'pull_request_target' || github.event_name == 'pull_request_review' || github.event_name == 'repository_dispatch' }}" "scheduler cancels stale PR/review/manual queue scans instead of accumulating merge/update attempts"
+	assert_file_contains "$workflow_file" "github.event_name == 'repository_dispatch' && github.event.client_payload.target_repository != '' && github.event.client_payload.pr_number != ''" "scheduler scopes targeted manual queue scans to the requested PR"
+	assert_file_contains "$workflow_file" "cancel-in-progress: \${{ github.event_name == 'pull_request_target' || github.event_name == 'pull_request_review' || github.event_name == 'repository_dispatch' || (github.event_name == 'workflow_run' && !github.event.workflow_run.pull_requests[0].number) }}" "scheduler cancels stale PR/review/manual queue scans instead of accumulating merge/update attempts"
 	assert_file_contains "$workflow_file" "timeout-minutes: 60" "organization sweep has enough headroom to finish the complete repository walk"
 	assert_file_contains "$workflow_file" "ORG_SWEEP_TRIGGER_REVIEWS: \${{ github.event_name == 'schedule' ||" "scheduled organization sweeps retry missing current-head OpenCode reviews"
 	assert_file_contains "$workflow_file" "ORG_SWEEP_ENABLE_AUTO_MERGE: \${{ github.event_name == 'schedule' ||" "scheduled organization sweeps merge approved current heads"

tests/test_opencode_review_normalize_output.py

@@ -2797,3 +2797,84 @@ def test_main_logs_the_exact_control_rejection_reason(tmp_path, capsys):
     stderr = capsys.readouterr().err
     assert "CONTROL_REJECTED candidate=1" in stderr
     assert "adversarial probe 1 evidence must state the observed proof result" in stderr
+
+
+def test_valid_control_repairs_trusted_model_probe_source_bindings(
+    tmp_path, monkeypatch
+):
+    """Substantive LLM probes survive formatting drift without weaker evidence."""
+    require_adversarial_validation(tmp_path, monkeypatch, "scripts/ci/example.py")
+    validation = adversarial_validation()
+    for probe in validation["probes"]:
+        probe["evidence"] = (
+            "Focused regression command passed and disproved the adversarial hypothesis. "
+            + source_line_receipt(f"line {probe['line']}")
+        )
+
+    normalized = norm.valid_control(
+        control(adversarial_validation=validation),
+        expected_head_sha="head",
+        expected_run_id="run",
+        expected_run_attempt="attempt",
+    )
+
+    assert normalized is not None
+    for index, probe in enumerate(normalized["adversarial_validation"]["probes"]):
+        line = 7 + index
+        assert f"scripts/ci/example.py:{line}" in probe["evidence"]
+        assert source_line_receipt(f"line {line}") in probe["evidence"]
+        assert probe["evidence"].count("source-line-sha256=") == 1
+
+
+def test_probe_binding_repair_does_not_invent_independent_observation(
+    tmp_path, monkeypatch
+):
+    """Canonical source binding cannot turn unsupported model prose into approval."""
+    require_adversarial_validation(tmp_path, monkeypatch, "scripts/ci/example.py")
+    validation = adversarial_validation()
+    for index, probe in enumerate(validation["probes"]):
+        probe["evidence"] = (
+            f"Description at scripts/ci/example.py:{7 + index} repeats the hypothesis. "
+            + source_line_receipt(f"line {7 + index}")
+        )
+    reasons = []
+
+    assert (
+        norm.valid_control(
+            control(adversarial_validation=validation),
+            expected_head_sha="head",
+            expected_run_id="run",
+            expected_run_attempt="attempt",
+            rejection_reasons=reasons,
+        )
+        is None
+    )
+    assert any("executed command" in reason for reason in reasons)
+
+
+@pytest.mark.parametrize(
+    "validation",
+    [
+        None,
+        {"probes": None},
+        {"probes": [None]},
+        {"probes": [{"path": 7, "line": 1, "evidence": "command passed"}]},
+        {"probes": [{"path": "scripts/ci/example.py", "line": True, "evidence": "command passed"}]},
+        {"probes": [{"path": "scripts/ci/example.py", "line": 0, "evidence": "command passed"}]},
+        {"probes": [{"path": "scripts/ci/example.py", "line": 7, "evidence": 9}]},
+        {"probes": [{"path": "missing.py", "line": 7, "evidence": "command passed"}]},
+        {
+            "probes": [
+                {
+                    "path": "../secrets/key.py",
+                    "line": 7,
+                    "evidence": "command passed",
+                }
+            ]
+        },
+    ],
+)
+def test_probe_binding_repair_preserves_unrepairable_shapes(validation):
+    """Malformed, unsafe, or unverifiable model evidence remains unchanged."""
+    candidate = control(adversarial_validation=validation)
+    assert norm.repair_adversarial_probe_source_bindings(candidate) is candidate