fix(OMN-8724): coerce dict payload into typed handler model at dispatch boundary + SkillRoutingError envelope

[jonahgabriel]

OMN-8724 — coerce dict payload into typed handler model at the dispatch boundary

Fixes the confirmed node_golden_chain_sweep crash via the canonical onex node path (BUG A in the integration-tooling triage): AttributeError: 'dict' object has no attribute 'chains' at handler_golden_chain_sweep.py because runtime_local_adapter._invoke_handle_method handed the raw decoded dict to handle(self, request: GoldenChainSweepRequest) instead of the typed model.

Layer decision — A1 (systemic), not A2 (narrow)

Chose A1: coerce the dict into the handler's declared input type inside omnibase_infra.runtime.runtime_local_adapter, fixing the whole dict-not-typed dispatch-boundary class (same family as OMN-13141 / the savings_estimation .get() bug) — not a per-node model_validate band-aid. Blast radius is bounded and regression-free: coercion fires only when (a) the payload is a raw dict AND (b) the handler's single positional parameter is annotated with a concrete BaseModel subclass. That combination previously always crashed, so the only behavior change is crash → success. All pre-existing pass-through paths (already-a-model, name-heuristic, kwargs-fanout, zero/var-keyword params) are unchanged.

Root cause completion — PEP 563 string annotations

Every node handler uses from __future__ import annotations, so inspect.signature(...).parameters[0].annotation was the literal string "GoldenChainSweepRequest", and the issubclass(annotation, BaseModel) check never matched. Fixed by resolving forward refs with inspect.signature(handle_method, eval_str=True) (graceful fallback to unevaluated on NameError).

Step 3 — SkillRoutingError envelope

cli_node default (non-receipt) mode now emits the documented SkillRoutingError JSON envelope to stdout on non-zero exit instead of leaving only a stderr traceback. The golden_chain_sweep SKILL.md documents this contract ("Non-zero exit emits a SkillRoutingError JSON envelope — surface it verbatim"). Envelope shape matches omnibase_core.cli.cli_run_node._emit_error. RuntimeLocal gains a public last_error property to populate message.

CI gate — real dispatch path

tests/unit/runtime/test_runtime_local_operation_match_typed_handler.py boots a typed single-model-parameter handler through RuntimeLocal.run_async() (the real operation_match dispatch path) and asserts the run COMPLETES. This is the test that "would have caught this" — handler-isolation tests pass while the live path crashed. Plus cli_node test asserting the SkillRoutingError envelope on a failed default-mode run.

Verification (local, in worktree)

• uv run ruff format / ruff check --fix on changed files — clean
• uv run mypy src/omnibase_infra/runtime/runtime_local_adapter.py src/omnibase_infra/runtime/runtime_local.py src/omnibase_infra/cli/cli_node.py --strict — Success, 3 files
• New test: FAILS (AttributeError: 'dict' object has no attribute 'correlation_id') on origin/dev, PASSES with fix
• uv run pytest tests/unit/runtime/ tests/unit/cli/ — 5157 passed (the only failures are an env-artifact skip-test that requires omnimarket installed, and a load-sensitive perf stress test that passes in isolation — both pre-existing, unrelated)
• pre-commit run on changed files — all hooks pass (incl. union budget gate)
• End-to-end proof: uv run onex node node_golden_chain_sweep → result=completed (was result=failed + traceback on origin/dev). .onex_state/workflow_result.json terminal_payload status: pass.

Companion PR

omnimarket jonah/omn-8724-golden-chain-sweep-downgrade-claim adds the runtime-injected correlation_id field to GoldenChainSweepRequest and downgrades the node/skill evidence claim (the node is pure compute over caller-supplied rows — never live evidence). Both land independently on dev.

dod_evidence / OCC pairing

Paired OCC receipt PR in onex_change_control carries contracts/OMN-8724.yaml + PASS receipts under drift/dod_receipts/OMN-8724/.

Evidence-Source: OCC#2626
Evidence-Ticket: OMN-8724

Closes OMN-8724 (crash + SkillRoutingError halves). Parent: OMN-8718.

[coderabbitai]

Warning

Review limit reached

@jonahgabriel, we couldn't start this review because you've reached your PR review rate limit.

More reviews will be available in 2 hours, 13 minutes, and 21 seconds. Learn how PR review limits work.

Your organization has used up its prepaid credits, and credit purchases are no longer available. Enable the review add-on in the billing tab to keep reviews running — you're only billed for reviews past your plan's rate limits ($0.25/file).

⌛ How to resolve this issue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

We recommend that you space out your commits to avoid hitting the rate limit.

🚦 How do rate limits work?

CodeRabbit enforces hourly rate limits for each developer per organization.

Our paid plans include higher PR review limits than trial, open-source, and free plans. In all cases, reviews become available again over time. During sustained high-volume PR review activity, CodeRabbit may temporarily slow when the next review becomes available.

Please see our Fair Usage Limits Policy for further information.

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 4b5ee97f-4f0d-4011-920c-692d23efb74f

📥 Commits

Reviewing files that changed from the base of the PR and between b56a3ff and 03712fb.

📒 Files selected for processing (5)

• src/omnibase_infra/cli/cli_node.py
• src/omnibase_infra/runtime/runtime_local.py
• src/omnibase_infra/runtime/runtime_local_adapter.py
• tests/unit/cli/test_cli_node.py
• tests/unit/runtime/test_runtime_local_operation_match_typed_handler.py

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch jonah/omn-8724-runtime-adapter-coerce-typed-model

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}