refactor: codec-owned provider-surface detection via a built-in registry

[zhongxuanwang-nv]

Overview

Implements RELAY-362, the follow-up requested in review of #291: move provider-surface detection out of the hard-coded if/else chains in codec::resolve into codec-owned descriptors iterated through a small built-in registry. resolve.rs is now provider-agnostic (no provider field-name literals), new built-in surfaces become additive, and a codec-owned provider hint can disambiguate the one ambiguous request shape (an Anthropic request without a top-level system). The change is additive, Rust-core-only, and behavior-preserving: every existing codec::resolve test passes unchanged and the adaptive path is untouched.

• I confirm this contribution is my own work, or I have the right to submit it under this project's license.
• I searched existing issues and open pull requests, and this does not duplicate existing work.

Details

• codec::resolve — added a pub(crate) SurfaceDescriptor (a hint-aware request detector, a response detector, and decode entry points) and a by-value static REGISTRY ordered by request-detection priority (Responses > Anthropic > Chat). The public functions are now provider-agnostic policy only: first registry match for requests, unique match for responses, fail-open normalize_*. ProviderSurface and every existing public signature are unchanged.
• New pub fn detect_request_surface_with_hint(body, Option<&str>) — detect_request_surface delegates to it with None for exact parity. A recognized hint can only upgrade the ambiguous messages-only shape; registry priority order keeps it from overriding a strong input/instructions/system signal.
• Codec-owned detection — each built-in codec module (openai_chat, openai_responses, anthropic) owns its detectors (as closures) and decode entry points in a SURFACE_DESCRIPTOR const. Only Anthropic's request detector is hint-aware (it claims a system-less messages body when the hint is "anthropic"); the OpenAI detectors ignore the hint. Adding a future built-in surface (e.g. Gemini) means registering a descriptor, not editing detection logic.
• Not wired into adaptive (deferred follow-up) — the hint is available but not yet called. Wiring it into resolve_request_surface / build_semantic_request_view would change ACG pass-through behavior, so it is intentionally left to a separate follow-up. The shape-only ACG path and its source-guard test are untouched.
• Tests — added hint coverage (parity with a None hint, the Anthropic upgrade, "never overrides a strong signal", non-object input); the existing detection/normalization tests are unchanged.

No breaking changes. Local validation: cargo test -p nemo-relay, -p nemo-relay-adaptive, and -p nemo-relay-pii-redaction pass; cargo clippy -p nemo-relay --all-features --all-targets, cargo fmt --check, and cargo doc -p nemo-relay are clean for the changed files. The full cross-language matrix is left to CI (no binding surfaces codec::resolve).

Where should the reviewer start?

crates/core/src/codec/resolve.rs — the SurfaceDescriptor + REGISTRY and the detection-policy functions (detect_request_surface_with_hint, detect_response_surface). Then the per-codec SURFACE_DESCRIPTOR blocks, especially anthropic.rs's hint-aware detect_request: registry priority order (Responses before Anthropic) is what keeps the "anthropic" hint from overriding a strong input/instructions signal.

Related Issues: (use one of the action keywords Closes / Fixes / Resolves / Relates to)

• Closes RELAY-362

Summary by CodeRabbit

• New Features

  • Added centralized, surface-descriptor-based request/response detection for Anthropic and OpenAI formats.
  • Introduced provider hints to disambiguate request classification when payload signals overlap.

• Bug Fixes

  • Reduced misclassification by selecting a format only when the response shape matches exactly one supported descriptor.
  • Updated normalization to decode via the detected format while preserving fail-open behavior on unrecognized inputs.

• Tests

  • Expanded unit coverage for hint behavior, request decoding, and stricter handling of ambiguous response shapes.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Enterprise

Run ID: 27add53a-10b9-4141-bb35-ef88f385a29f

📥 Commits

Reviewing files that changed from the base of the PR and between ac4eac2 and 7bb3cb6.

📒 Files selected for processing (2)

• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

📜 Recent review details

⏰ Context from checks skipped due to timeout. (2)

• GitHub Check: Check / Run
• GitHub Check: Preview docs

🧰 Additional context used

📓 Path-based instructions (14)

**/*.rs

📄 CodeRabbit inference engine (.agents/skills/add-binding-feature/SKILL.md)

Use snake_case naming convention for Rust identifiers (e.g., nemo_relay_tool_call)

**/*.rs: Any Rust change must run just test-rust
Any Rust change must run cargo fmt --all
Any Rust change must run cargo clippy --workspace --all-targets -- -D warnings

**/*.rs: Run cargo fmt --all for all FFI work since it is Rust work
Run just test-rust to validate FFI changes
Run cargo clippy --workspace --all-targets -- -D warnings to enforce strict linting on FFI work

When Rust files changed as part of Go work, also run cargo fmt --all, just test-rust, and cargo clippy --workspace --all-targets -- -D warnings

**/*.rs: Run cargo fmt --all when Rust files are changed as part of Node work
Run cargo clippy --workspace --all-targets -- -D warnings when Rust files are changed as part of Node work
Run just test-rust when Rust files are changed as part of Node work

**/*.rs: Run cargo fmt --all to format all Rust code
Run cargo clippy --workspace --all-targets -- -D warnings to enforce all clippy lints as errors

**/*.rs: Run cargo fmt --all when Rust files changed as part of WebAssembly work
Run cargo clippy --workspace --all-targets -- -D warnings when Rust files changed as part of WebAssembly work

**/*.rs: If any Rust code changed, always run just test-rust
If any Rust code changed, also run cargo fmt --all
If any Rust code changed, also run cargo clippy --workspace --all-targets -- -D warnings
Run Rust formatting with cargo fmt --all
Run Rust linting with cargo clippy --workspace --all-targets -- -D warnings

**/*.rs: Use cargo fmt for Rust code formatting
Run cargo clippy -- -D warnings to lint Rust code and treat all warnings as errors
Use Rust snake_case naming convention for Rust identifiers
Include SPDX license header in all Rust source files using double-slash comment syntax
Validate Rust code with uv run pre-commit run --all-files to enforce cargo fmt formatting check, cargo clippy lints, and cargo deny aud...

Files:

• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

**/{Cargo.toml,**/*.rs}

📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)

Maintain consistency between Rust package names in Cargo.toml and their actual usage across the codebase

Files:

• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

**/*.{h,hpp,c,cpp,rs}

📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)

Ensure FFI header and library naming follows consistent conventions across platform-specific builds

Files:

• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

{crates/core,crates/adaptive}/**/*

📄 CodeRabbit inference engine (.agents/skills/prepare-pr/SKILL.md)

Changes to crates/core or crates/adaptive must run the full language matrix

Files:

• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

**/*.{rs,toml}

📄 CodeRabbit inference engine (.agents/skills/rename-surfaces/SKILL.md)

Update Rust crate names and module prefixes during coordinated rename operations

Files:

• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

crates/core/**/*.rs

📄 CodeRabbit inference engine (.agents/skills/test-go-binding/SKILL.md)

If the change touched crates/core or shared runtime semantics, also use validate-change for broader validation

crates/core/**/*.rs: Use Json = serde_json::Value in Rust-facing runtime APIs where the existing code expects JSON payloads.
Use Result<T> with FlowError in core runtime paths. Keep errors explicit and binding-appropriate at the wrapper layer.

Files:

• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

crates/{core,adaptive}/**

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

If crates/core or crates/adaptive changed, run the full matrix across Rust, Python, Go, Node.js, and WebAssembly

Files:

• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

**/*.{rs,py,js,ts,tsx,jsx,go,sh,toml,yaml,yml,md}

📄 CodeRabbit inference engine (AGENTS.md)

Keep SPDX headers on source, docs, scripts, and configuration files. The project is Apache-2.0.

Files:

• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

**/*.{rs,py,go,js,ts,tsx}

📄 CodeRabbit inference engine (AGENTS.md)

Follow binding naming conventions: Rust and Python use snake_case, C FFI exports prefixed nemo_relay_, Go uses PascalCase for public APIs, Node.js uses camelCase.

Files:

• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

crates/**/*.rs

📄 CodeRabbit inference engine (AGENTS.md)

crates/**/*.rs: Keep async behavior on the existing tokio-based model. Bindings should preserve callback and future lifetimes rather than blocking or hiding async work unexpectedly.
Use Json = serde_json::Value in Rust-facing runtime APIs for JSON payload handling.

Files:

• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

**

⚙️ CodeRabbit configuration file

**:

AGENTS.md

This file provides guidance to agents, including Claude Code and OpenAI Codex, when working in this repository.

Project Overview

NeMo Relay is a multi-language agent runtime framework for execution scopes, lifecycle events, middleware, plugins, and observability around tool and LLM calls. The core runtime is Rust. Primary supported bindings are Rust, Python, and Node.js. Go, WebAssembly, and the raw C FFI are experimental and source-first.

The shared runtime model is:

1. Scope stacks decide where work belongs and which scope-local behavior is visible.
2. Middleware registries decide what guardrails and intercepts run around managed calls.
3. Plugins install reusable runtime behavior from configuration.
4. Events record runtime behavior in ATOF form.
5. Subscribers and exporters consume events in-process or export them to ATIF, OpenTelemetry, OpenInference, or other backends.

Repository Structure

The repository layout separates the Rust runtime, language bindings, documentation,
integration patches, and agent-facing skills.

crates/
  core/       # Rust core runtime crate, published as nemo-relay
  adaptive/   # Adaptive runtime primitives and plugin components
  python/     # PyO3 native extension for the Python package
  ffi/        # Raw C ABI layer used by downstream bindings such as Go
  node/       # NAPI Node.js binding and JavaScript/TypeScript entry points
  wasm/       # wasm-bindgen WebAssembly binding and JS wrappers
python/
  nemo_relay/  # Python wrapper package: scopes, tools, LLM, middleware, typed helpers, plugins, adaptive helpers
  tests/      # Python tests
go/
  nemo_relay/  # Experimental Go CGo binding and tests
fern/         # Fern documentation site
scripts/      # Stable wrappers and helper scripts; build/test/docs entry points live in justfile
third_party/  # P...

Files:

• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

crates/{core,adaptive}/**/*.rs

⚙️ CodeRabbit configuration file

crates/{core,adaptive}/**/*.rs: Review the Rust runtime for async correctness, scope isolation, middleware ordering, and event lifecycle regressions.
Pay close attention to task-local/thread-local scope propagation, callback lifetimes, stream finalization, and root_uuid isolation.
Public API changes should preserve existing behavior unless tests and docs show the intended migration path.

Files:

• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

{crates/adaptive/**/*.rs,**/*test*.{rs,py,go,ts,js},**/*adaptive*test*.{rs,py,go,ts,js},docs/plugins/adaptive/**}

📄 CodeRabbit inference engine (.agents/skills/maintain-optimizer/SKILL.md)

Maintain documented and tested validation and report behavior for adaptive surfaces

Files:

• crates/core/tests/unit/codec/resolve_tests.rs

{crates/**/tests/**,python/tests/**,go/nemo_relay/**/*_test.go}

⚙️ CodeRabbit configuration file

{crates/**/tests/**,python/tests/**,go/nemo_relay/**/*_test.go}: Tests should cover the behavior promised by the changed API surface, including error paths and cross-request isolation where relevant.
Prefer assertions on lifecycle events, scope stacks, middleware ordering, and binding parity over shallow smoke tests.

Files:

• crates/core/tests/unit/codec/resolve_tests.rs

🔇 Additional comments (2)

crates/core/src/codec/resolve.rs (1)

63-65: LGTM!

crates/core/tests/unit/codec/resolve_tests.rs (1)

221-230: LGTM!

---

Walkthrough

The PR adds registry-backed surface descriptors for OpenAI Chat, OpenAI Responses, and Anthropic Messages, routes request/response normalization through the registry, and expands tests for hint-aware request detection and ambiguous response shapes.

Changes

Registry-backed codec surface detection

Layer / File(s)	Summary
Registry and normalization flow crates/core/src/codec/resolve.rs	SurfaceDescriptor and detector plumbing are added, request detection accepts an optional provider hint, response detection requires exactly one matching descriptor, and normalization decodes through the selected descriptor.
Built-in codec descriptors crates/core/src/codec/openai_responses.rs, crates/core/src/codec/anthropic.rs, crates/core/src/codec/openai_chat.rs	OpenAI Responses, Anthropic Messages, and OpenAI Chat each add a SURFACE_DESCRIPTOR with JSON shape predicates and decode routing to the codec implementation.
Hint-aware detection tests crates/core/tests/unit/codec/resolve_tests.rs	Unit tests cover detect_request_surface_with_hint across matching None, provider-specific hints, strong signals, non-object or keyless bodies, and ambiguous response-shape handling.

Estimated review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

---

Important

Pre-merge checks failed

Please resolve all errors before merging. Addressing warnings is optional.

❌ Failed checks (1 error)

Check name	Status	Explanation	Resolution
Title check	❌ Error	The title is relevant, but it is 72 characters long, which exceeds the under-72-character limit.	Shorten the title to 71 characters or fewer while keeping the Conventional Commits prefix.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description check	✅ Passed	The description follows the template with Overview, Details, Reviewer Start, checkboxes, and a related issue reference.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
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.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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

[github-actions]

Fern docs preview: https://nvidia-preview-pull-request-301.docs.buildwithfern.com/nemo/relay (​https://nvidia-preview-pull-request-301.docs.buildwithfern.com/nemo/relay​)

[zhongxuanwang-nv]

@coderabbitai 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.

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@crates/core/src/codec/resolve.rs`:
- Around line 82-106: The exact-one response descriptor selection rule is
duplicated between detect_response_surface and normalize_response, so extract
that matching logic into a single private helper and have both functions use it.
Keep the helper responsible for iterating REGISTRY and returning the sole
matching descriptor only when there is exactly one match, then reuse it in both
detect_response_surface and normalize_response so any future change to matching
behavior stays consistent.

In `@crates/core/tests/unit/codec/resolve_tests.rs`:
- Around line 269-289: Add a response-side unit test for the “exactly one match”
behavior in detect_response_surface/normalize_response: create cases where a
response body matches zero detectors and where it matches two or more detectors,
and assert the result is None. Place the test near
hint_never_overrides_strong_signals so the new contract is covered alongside the
existing resolve_tests helpers and detector logic.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Enterprise

Run ID: 3bb781d9-a210-4683-bd6e-49bb1898f05d

📥 Commits

Reviewing files that changed from the base of the PR and between 6b1ce42 and 261fc7e.

📒 Files selected for processing (5)

• crates/core/src/codec/anthropic.rs
• crates/core/src/codec/openai_chat.rs
• crates/core/src/codec/openai_responses.rs
• crates/core/src/codec/resolve.rs
• crates/core/tests/unit/codec/resolve_tests.rs

📜 Review details

⏰ Context from checks skipped due to timeout. (2)

• GitHub Check: Check / Run
• GitHub Check: Preview docs

🧰 Additional context used

📓 Path-based instructions (14)

**/*.rs

📄 CodeRabbit inference engine (.agents/skills/add-binding-feature/SKILL.md)

Use snake_case naming convention for Rust identifiers (e.g., nemo_relay_tool_call)

**/*.rs: Any Rust change must run just test-rust
Any Rust change must run cargo fmt --all
Any Rust change must run cargo clippy --workspace --all-targets -- -D warnings

**/*.rs: Run cargo fmt --all for all FFI work since it is Rust work
Run just test-rust to validate FFI changes
Run cargo clippy --workspace --all-targets -- -D warnings to enforce strict linting on FFI work

When Rust files changed as part of Go work, also run cargo fmt --all, just test-rust, and cargo clippy --workspace --all-targets -- -D warnings

**/*.rs: Run cargo fmt --all when Rust files are changed as part of Node work
Run cargo clippy --workspace --all-targets -- -D warnings when Rust files are changed as part of Node work
Run just test-rust when Rust files are changed as part of Node work

**/*.rs: Run cargo fmt --all to format all Rust code
Run cargo clippy --workspace --all-targets -- -D warnings to enforce all clippy lints as errors

**/*.rs: Run cargo fmt --all when Rust files changed as part of WebAssembly work
Run cargo clippy --workspace --all-targets -- -D warnings when Rust files changed as part of WebAssembly work

**/*.rs: If any Rust code changed, always run just test-rust
If any Rust code changed, also run cargo fmt --all
If any Rust code changed, also run cargo clippy --workspace --all-targets -- -D warnings
Run Rust formatting with cargo fmt --all
Run Rust linting with cargo clippy --workspace --all-targets -- -D warnings

**/*.rs: Use cargo fmt for Rust code formatting
Run cargo clippy -- -D warnings to lint Rust code and treat all warnings as errors
Use Rust snake_case naming convention for Rust identifiers
Include SPDX license header in all Rust source files using double-slash comment syntax
Validate Rust code with uv run pre-commit run --all-files to enforce cargo fmt formatting check, cargo clippy lints, and cargo deny aud...

Files:

• crates/core/src/codec/openai_responses.rs
• crates/core/src/codec/openai_chat.rs
• crates/core/tests/unit/codec/resolve_tests.rs
• crates/core/src/codec/resolve.rs
• crates/core/src/codec/anthropic.rs

**/{Cargo.toml,**/*.rs}

📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)

Maintain consistency between Rust package names in Cargo.toml and their actual usage across the codebase

Files:

• crates/core/src/codec/openai_responses.rs
• crates/core/src/codec/openai_chat.rs
• crates/core/tests/unit/codec/resolve_tests.rs
• crates/core/src/codec/resolve.rs
• crates/core/src/codec/anthropic.rs

**/*.{h,hpp,c,cpp,rs}

📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)

Ensure FFI header and library naming follows consistent conventions across platform-specific builds

Files:

• crates/core/src/codec/openai_responses.rs
• crates/core/src/codec/openai_chat.rs
• crates/core/tests/unit/codec/resolve_tests.rs
• crates/core/src/codec/resolve.rs
• crates/core/src/codec/anthropic.rs

{crates/core,crates/adaptive}/**/*

📄 CodeRabbit inference engine (.agents/skills/prepare-pr/SKILL.md)

Changes to crates/core or crates/adaptive must run the full language matrix

Files:

• crates/core/src/codec/openai_responses.rs
• crates/core/src/codec/openai_chat.rs
• crates/core/tests/unit/codec/resolve_tests.rs
• crates/core/src/codec/resolve.rs
• crates/core/src/codec/anthropic.rs

**/*.{rs,toml}

📄 CodeRabbit inference engine (.agents/skills/rename-surfaces/SKILL.md)

Update Rust crate names and module prefixes during coordinated rename operations

Files:

• crates/core/src/codec/openai_responses.rs
• crates/core/src/codec/openai_chat.rs
• crates/core/tests/unit/codec/resolve_tests.rs
• crates/core/src/codec/resolve.rs
• crates/core/src/codec/anthropic.rs

crates/core/**/*.rs

📄 CodeRabbit inference engine (.agents/skills/test-go-binding/SKILL.md)

If the change touched crates/core or shared runtime semantics, also use validate-change for broader validation

crates/core/**/*.rs: Use Json = serde_json::Value in Rust-facing runtime APIs where the existing code expects JSON payloads.
Use Result<T> with FlowError in core runtime paths. Keep errors explicit and binding-appropriate at the wrapper layer.

Files:

• crates/core/src/codec/openai_responses.rs
• crates/core/src/codec/openai_chat.rs
• crates/core/tests/unit/codec/resolve_tests.rs
• crates/core/src/codec/resolve.rs
• crates/core/src/codec/anthropic.rs

crates/{core,adaptive}/**

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

If crates/core or crates/adaptive changed, run the full matrix across Rust, Python, Go, Node.js, and WebAssembly

Files:

• crates/core/src/codec/openai_responses.rs
• crates/core/src/codec/openai_chat.rs
• crates/core/tests/unit/codec/resolve_tests.rs
• crates/core/src/codec/resolve.rs
• crates/core/src/codec/anthropic.rs

**/*.{rs,py,js,ts,tsx,jsx,go,sh,toml,yaml,yml,md}

📄 CodeRabbit inference engine (AGENTS.md)

Keep SPDX headers on source, docs, scripts, and configuration files. The project is Apache-2.0.

Files:

• crates/core/src/codec/openai_responses.rs
• crates/core/src/codec/openai_chat.rs
• crates/core/tests/unit/codec/resolve_tests.rs
• crates/core/src/codec/resolve.rs
• crates/core/src/codec/anthropic.rs

**/*.{rs,py,go,js,ts,tsx}

📄 CodeRabbit inference engine (AGENTS.md)

Follow binding naming conventions: Rust and Python use snake_case, C FFI exports prefixed nemo_relay_, Go uses PascalCase for public APIs, Node.js uses camelCase.

Files:

• crates/core/src/codec/openai_responses.rs
• crates/core/src/codec/openai_chat.rs
• crates/core/tests/unit/codec/resolve_tests.rs
• crates/core/src/codec/resolve.rs
• crates/core/src/codec/anthropic.rs

crates/**/*.rs

📄 CodeRabbit inference engine (AGENTS.md)

crates/**/*.rs: Keep async behavior on the existing tokio-based model. Bindings should preserve callback and future lifetimes rather than blocking or hiding async work unexpectedly.
Use Json = serde_json::Value in Rust-facing runtime APIs for JSON payload handling.

Files:

• crates/core/src/codec/openai_responses.rs
• crates/core/src/codec/openai_chat.rs
• crates/core/tests/unit/codec/resolve_tests.rs
• crates/core/src/codec/resolve.rs
• crates/core/src/codec/anthropic.rs

**

⚙️ CodeRabbit configuration file

**:

AGENTS.md

This file provides guidance to agents, including Claude Code and OpenAI Codex, when working in this repository.

Project Overview

NeMo Relay is a multi-language agent runtime framework for execution scopes, lifecycle events, middleware, plugins, and observability around tool and LLM calls. The core runtime is Rust. Primary supported bindings are Rust, Python, and Node.js. Go, WebAssembly, and the raw C FFI are experimental and source-first.

The shared runtime model is:

1. Scope stacks decide where work belongs and which scope-local behavior is visible.
2. Middleware registries decide what guardrails and intercepts run around managed calls.
3. Plugins install reusable runtime behavior from configuration.
4. Events record runtime behavior in ATOF form.
5. Subscribers and exporters consume events in-process or export them to ATIF, OpenTelemetry, OpenInference, or other backends.

Repository Structure

The repository layout separates the Rust runtime, language bindings, documentation,
integration patches, and agent-facing skills.

crates/
  core/       # Rust core runtime crate, published as nemo-relay
  adaptive/   # Adaptive runtime primitives and plugin components
  python/     # PyO3 native extension for the Python package
  ffi/        # Raw C ABI layer used by downstream bindings such as Go
  node/       # NAPI Node.js binding and JavaScript/TypeScript entry points
  wasm/       # wasm-bindgen WebAssembly binding and JS wrappers
python/
  nemo_relay/  # Python wrapper package: scopes, tools, LLM, middleware, typed helpers, plugins, adaptive helpers
  tests/      # Python tests
go/
  nemo_relay/  # Experimental Go CGo binding and tests
fern/         # Fern documentation site
scripts/      # Stable wrappers and helper scripts; build/test/docs entry points live in justfile
third_party/  # P...

Files:

• crates/core/src/codec/openai_responses.rs
• crates/core/src/codec/openai_chat.rs
• crates/core/tests/unit/codec/resolve_tests.rs
• crates/core/src/codec/resolve.rs
• crates/core/src/codec/anthropic.rs

crates/{core,adaptive}/**/*.rs

⚙️ CodeRabbit configuration file

crates/{core,adaptive}/**/*.rs: Review the Rust runtime for async correctness, scope isolation, middleware ordering, and event lifecycle regressions.
Pay close attention to task-local/thread-local scope propagation, callback lifetimes, stream finalization, and root_uuid isolation.
Public API changes should preserve existing behavior unless tests and docs show the intended migration path.

Files:

• crates/core/src/codec/openai_responses.rs
• crates/core/src/codec/openai_chat.rs
• crates/core/tests/unit/codec/resolve_tests.rs
• crates/core/src/codec/resolve.rs
• crates/core/src/codec/anthropic.rs

{crates/adaptive/**/*.rs,**/*test*.{rs,py,go,ts,js},**/*adaptive*test*.{rs,py,go,ts,js},docs/plugins/adaptive/**}

📄 CodeRabbit inference engine (.agents/skills/maintain-optimizer/SKILL.md)

Maintain documented and tested validation and report behavior for adaptive surfaces

Files:

• crates/core/tests/unit/codec/resolve_tests.rs

{crates/**/tests/**,python/tests/**,go/nemo_relay/**/*_test.go}

⚙️ CodeRabbit configuration file

{crates/**/tests/**,python/tests/**,go/nemo_relay/**/*_test.go}: Tests should cover the behavior promised by the changed API surface, including error paths and cross-request isolation where relevant.
Prefer assertions on lifecycle events, scope stacks, middleware ordering, and binding parity over shallow smoke tests.

Files:

• crates/core/tests/unit/codec/resolve_tests.rs

🔇 Additional comments (6)

crates/core/src/codec/resolve.rs (2)

27-48: LGTM!

---

58-73: LGTM!

crates/core/src/codec/openai_responses.rs (1)

46-55: LGTM!

crates/core/src/codec/anthropic.rs (1)

47-60: LGTM!

crates/core/src/codec/openai_chat.rs (1)

34-40: LGTM!

crates/core/tests/unit/codec/resolve_tests.rs (1)

226-301: LGTM!

[mnajafian-nv]

LGTM, great work! This is a conditional approval pending Green CI. I did a focused pass on scope fit, resolver correctness, and regression risk. The registry refactor looks clean and behavior-preserving to me overall; I left one small API-doc clarity note for the new public hint entry point.

[zhongxuanwang-nv]

/ok to test 7bb3cb6

[mnajafian-nv]

Thank you, LGTM!

[zhongxuanwang-nv]

/merge