fix(ci): allow bounded review request envelopes

docs/doctoring/contextual-orchestrator-vendored-sidecar.md

@@ -7,6 +7,17 @@
   prioritized.
 - **Decision record:** [`docs/adr/0003-contextual-orchestrator-vendored-free-zdr.md`](adr/0003-contextual-orchestrator-vendored-free-zdr.md)
 
+## 2026-08-28 Strix request envelope
+
+The first post-#1373 main Strix execution reached the gateway and selected
+`openai/orchestrator/free`, but the pinned server rejected the initial agent
+request with HTTP 413 (`request_too_large`). The vendored server's generic
+`SecurityConfig.max_body_bytes` default is 64 KiB; Strix and Noema requests
+include tool schemas and repository context and therefore need a larger,
+still-bounded integration envelope. The review launcher now sets an explicit
+8 MiB limit for this sidecar only; the library default remains unchanged for
+other deployments. A 413 remains fail-closed if a request exceeds that bound.
+
 ## What changed
 
 `pr-review-autofix.yml` now provisions the sidecar

docs/product-technical-gap-baseline.md

@@ -240,8 +240,9 @@ flowchart LR
 
 ## 2026-08-28 current-main routing and runtime recheck
 
-- Current protected main is `24ee38b097dbfc1a895e1199ade48cff36431d05`,
-  the merge commit for #1370. #1364 is merged at
+- Current protected main is `8f84b661e468de451ba5c076dc938f342bf52d70`,
+  the merge commit for #1373 (following #1370 at
+  `24ee38b097dbfc1a895e1199ade48cff36431d05`). #1364 is merged at
   `f8823a544c3c4c046977f8511f683e85f83eb496`; #1360 is merged at
   `17052a7ca3c16db90932a4d6036b43165ddee418`.
 - The current Required OpenCode dispatch, `noema-review.yml`, `strix.yml`,
@@ -287,6 +288,41 @@ flowchart LR
   Strix completion and an independently authorized Noema verdict remain
   separate evidence items.
 
+## 2026-08-28 post-#1373 request-envelope recheck
+
+- #1373 was merged by `seonghobae` at `8f84b661e468de451ba5c076dc938f342bf52d70`
+  to exercise the post-merge runtime path. Main Strix run `33143805461`
+  reached the contextual-orchestrator sidecar and sent the qualified
+  `openai/orchestrator/free` request, then failed closed with HTTP 413
+  `request_too_large` from the pinned gateway. This proves the earlier model
+  qualification defect was repaired, but the review request envelope was
+  still smaller than the Strix/Noema tool-and-source context.
+- The fix is scoped to the review launcher: use an explicit bounded 8 MiB
+  `SecurityConfig.max_body_bytes` for the sidecar while preserving the
+  contextual-orchestrator library's generic 64 KiB default. Noema run
+  `33143860315` was a successful `workflow_run` event handler but skipped
+  because the push event had no associated pull request; it is not an LLM
+  verdict.
+
+## 2026-08-28 #1374 trusted-base runtime boundary
+
+- Follow-up PR #1374 is open at head
+  `d3e7cee4b01219cb0a93f1a4249049de3bf05b4b`, based on current main
+  `8f84b661e468de451ba5c076dc938f342bf52d70`. Its launcher sets the bounded
+  8 MiB review envelope, and its sidecar boot check validates that keyword
+  against the exact pinned orchestrator SHA before discovery.
+- PR-target Strix run `33145070402` used trusted workflow source SHA
+  `8f84b661e468de451ba5c076dc938f342bf52d70`, not the PR launcher. It reached
+  the pinned sidecar and then failed three bounded attempts with HTTP 413
+  `request_too_large`; this is evidence of the pre-merge trusted-base path,
+  not evidence that #1374's launcher setting failed.
+- PR-target Noema run `33145070347` also reached the pinned sidecar and set
+  `orchestrator/free`, then skipped before the LLM call because the current
+  head had no primary OpenCode approval. Required OpenCode run `33145070315`
+  failed closed for the same missing current-head verdict. Therefore the
+  envelope fix still needs a normal governed merge followed by a post-merge
+  Strix runtime result; no protected completion is claimed here.
+
 ## 5. 실행 루프와 고객의 다음 행동
 
 각 hourly pass는 아래 순서를 유지한다.

scripts/ci/contextual_orchestrator_review_launcher.py

@@ -27,6 +27,12 @@
 from pathlib import Path
 
 
+# The vendored server's generic 64 KiB default is intentionally conservative,
+# but Strix and Noema send tool schemas plus repository context in one request.
+# Keep the review-specific envelope bounded without weakening the library default.
+REVIEW_MAX_BODY_BYTES = 8 * 1024 * 1024
+
+
 def _free_report_rows(discovered: list[object]) -> list[dict[str, object]]:
     """Convert in-process discovered models into free-only report rows.
 
@@ -157,7 +163,10 @@ def main(argv: list[str] | None = None) -> int:
         orchestrator,
         host=args.host,
         port=args.port,
-        security=SecurityConfig(auth_token=auth_token),
+        security=SecurityConfig(
+            auth_token=auth_token,
+            max_body_bytes=REVIEW_MAX_BODY_BYTES,
+        ),
     )
     return 0
 

scripts/ci/contextual_orchestrator_review_sidecar.sh

@@ -74,6 +74,8 @@ python3 -m pip install --quiet --disable-pip-version-check --no-cache-dir \
   -r "$requirements_lock"
 PYTHONPATH="$ORCHESTRATOR_SOURCE:$ORG_REPO_ROOT" "$(command -v python3)" -c \
   'from contextual_orchestrator.credentials import get_credential; from contextual_orchestrator.model_discovery import discover_all_models, free_discovered_models; from contextual_orchestrator.orchestrator import ModelClient, TaskOrchestrator, load_agents; from contextual_orchestrator.review_gateway import register_review_credentials; from contextual_orchestrator.server import SecurityConfig, serve'
+PYTHONPATH="$ORCHESTRATOR_SOURCE:$ORG_REPO_ROOT" "$(command -v python3)" -c \
+  'from contextual_orchestrator.server import SecurityConfig; from scripts.ci.contextual_orchestrator_review_launcher import REVIEW_MAX_BODY_BYTES; SecurityConfig(auth_token="contract", max_body_bytes=REVIEW_MAX_BODY_BYTES)'
 
 discovery_report="$ORCHESTRATOR_WORK/discovery-free.json"
 zdr_feed="$ORCHESTRATOR_WORK/openrouter-zdr-endpoints.json"

tests/test_contextual_orchestrator_review_sidecar_contract.py

@@ -129,6 +129,19 @@ def test_launcher_requires_gateway_token_and_a_provider_credential() -> None:
     assert "requires at least one provider credential in the KV" in text
 
 
+def test_launcher_sets_a_bounded_review_request_body_limit() -> None:
+    """Large review envelopes fit without changing the library's generic default."""
+    text = _read(LAUNCHER)
+    assert "REVIEW_MAX_BODY_BYTES = 8 * 1024 * 1024" in text
+    assert "max_body_bytes=REVIEW_MAX_BODY_BYTES" in text
+
+
+def test_sidecar_validates_the_pinned_server_body_limit_constructor() -> None:
+    """The exact vendored SHA must accept the review envelope keyword at boot."""
+    text = _read(SIDECAR)
+    assert 'from contextual_orchestrator.server import SecurityConfig; from scripts.ci.contextual_orchestrator_review_launcher import REVIEW_MAX_BODY_BYTES; SecurityConfig(auth_token="contract", max_body_bytes=REVIEW_MAX_BODY_BYTES)' in text
+
+
 def test_autofix_workflow_provisions_sidecar_with_all_five_secrets() -> None:
     """The write-capable autofix path bootstraps the gateway with the five keys."""
     workflow = _read(AUTOFIX_WORKFLOW)