fix: Stamp the selected ModelId on LlmRequest much earlier

[grahamking]

Previously libsy would give CallModel a request with the incoming model
ID (e.g. switchyard/random), and a Decision with the correct id (e.g. openai/gpt-oss-120b)
as chosen by the routing algorithm. libsy-llm-client would stamp the later onto
the former before making the actual call.

Now Driver::call_model stamps the correct model ID onto LlmRequest,
so everything after that can use LlmRequest directly.

This was confusing me for a while. Seems easier to understand like this.

This also means CallModel can read the selected_model_id() off the
request now, which gives us options for changing Decision later.

The RoutedLlmClient::call method took a Decision purely to extract it's
model_id and stamp it on the request, so we also simplify that to call(Request).

Claude's review:

The net effect is simpler client code and a cleaner contract on CallModel.

Assisted-by: Codex:GPT 5.6 Sol high
Signed-off-by: Graham King grahamk@nvidia.com

Summary by CodeRabbit

• Improvements

  • Requests now consistently use the model selected by routing.
  • Random-routing behavior now preserves the selected model in outgoing client requests.
  • Model selection and request handling are more consistent across token counting, rewrites, and standard calls.

• Documentation

  • Clarified how selected models are represented and propagated through requests and routed clients.

[grahamking]

@CodeRabbit review

[coderabbitai]

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

[github-actions]

PR Preview Action v1.8.1
🚀 View preview at https://NVIDIA-NeMo.github.io/Switchyard/pr-preview/pr-393/
Built to branch gh-pages at 2026-08-13 14:45 UTC. Preview will be ready when the GitHub Pages deployment is complete.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: fe914580-0321-44af-b938-58e6562f8186

📥 Commits

Reviewing files that changed from the base of the PR and between 0151621 and aa06e52.

📒 Files selected for processing (7)

• crates/libsy-llm-client/src/client.rs
• crates/libsy-llm-client/src/run.rs
• crates/libsy/src/core/algorithm.rs
• crates/protocol/src/client.rs
• crates/protocol/src/envelope.rs
• crates/protocol/src/llm.rs
• tests/test_libsy_minimal_bindings.py

---

Walkthrough

Routing now stamps the selected model into requests. LLM clients, tracing, error handling, tests, and documentation use the request model as the routed target.

Changes

Request-Model Routing

Layer / File(s)	Summary
Stamp the selected model crates/libsy/src/core/algorithm.rs	Driver::call_model writes the selected model into request.llm_request.model. CallModel::selected_model_id() exposes the stamped value. Related tests verify the request and accessor.
Dispatch the request model crates/libsy-llm-client/src/client.rs, crates/protocol/src/*.rs	Client entry points resolve models into request payloads. RoutedLlmClient::call forwards the request model without overriding it. Protocol documentation describes the updated model contract.
Use the selected model in observations crates/libsy-llm-client/src/run.rs, tests/test_libsy_minimal_bindings.py	Tracing, errors, and call observations use CallModel::selected_model_id(). The bindings test verifies that the selected model reaches the client request.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Mergeability Score: ⚪ Minimal · up to aa06e

The change is merge-ready after normal checks and review; no actionable merge-blocking risk remains.

Poem

I stamped the model, said the rabbit with cheer,
Then sent it downstream, precise and clear.
The client now follows the request’s bright trail,
While traces and tests confirm the tale.
“Hop!” said the code, “the target is here!”

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the main change: stamping the selected model ID on LlmRequest earlier.

---

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

[ayushag-nv]

Nice Catch. Looks good.