RHIDP-14130: OKP experience improvements - citation URLs, search mode, ogx provider migration

[maysunfaisal]

Description

Fixes and enhances OKP (Offline Knowledge Portal) integration across both rag.tool and rag.inline paths, and adds air-gap support via BM25 keyword search.

What changed

1. Fix OKP citation URLs for rag.tool path

Previously, OKP citation URLs only worked in the rag.inline path. The rag.tool path (Responses API / file_search) was broken because:

• The tool path looked for doc_url in result attributes, but Solr provider returns reference_url and source_path instead
• No OKP-specific URL construction existed in the tool path — reference_url is origin-relative, not a full URL

Fix: Added OKP URL building to both rag.tool (src/utils/responses.py, src/utils/agents/tool_processor.py) and rag.inline paths, supporting both offline (localhost) and online (docs.redhat.com) modes.

2. Fix lexical → keyword search mode translation

The Solr provider internally used the term lexical for BM25 search, but LCORE config exposed it as keyword (more user-friendly). This mismatch meant BM25 mode was silently broken — requests with search_mode: keyword would not translate correctly to the Solr provider.

Fix: Added mode translation in src/utils/vector_search.py so keyword maps correctly to the Solr provider's lexical mode.

3. Add configurable search_mode for air-gap deployments

In air-gapped environments, users cannot download embedding models from HuggingFace. Previously there was no way to configure the search mode — it always defaulted to vector/hybrid search which requires an embedding model.

Fix: Added search_mode field to OkpConfiguration with three options:

• vector (default) — semantic search, requires embedding model
• keyword — BM25 text search, no embedding model needed (air-gap friendly)
• hybrid — combines both

Also added per-request search_mode override in query request models.

4. Fix Solr embedding model registration prefix

The Solr embedding model was registered as solr_embedding, but the sentence-transformers inference provider expects the sentence-transformers/ prefix.

Fix: Added SOLR_EMBEDDING_MODEL_ID constant (sentence-transformers/solr_embedding) in src/constants.py and used it consistently in src/llama_stack_configuration.py for model and vector store registration.

5. Update providers submodule + Containerfile pin

Updated to include merged lightspeed-providers PR #160 which migrates the Solr vector_io provider from llama_stack to ogx imports (faf6a89).

6. Regenerated OpenAPI schema + updated tests

• Regenerated docs/devel_doc/openapi.json to include new search_mode field
• Updated 10 config dump test assertions and Solr enrichment test for new model ID prefix

7. Propagate search_mode to OGX for rag.tool keyword search

Previously, search_mode: keyword only affected the rag.inline path. The rag.tool path (file_search via Responses API) always defaulted to vector similarity search because OGX's default_search_mode config was never set by LCORE.

Fix: enrich_solr() now reads search_mode from the OKP config and sets vector_stores.chunk_retrieval_params.default_search_mode in the OGX top-level config. OGX auto-injects this into both the file_search tool runtime and responses provider, so both rag paths now respect search_mode: keyword. This is critical for air-gap environments where no embedding model is available.

Air-gap scenario

With search_mode: keyword, OKP uses Solr BM25 text search — no embedding model is downloaded or needed. This works for both rag.tool and rag.inline paths:

# rag.tool — air-gap with keyword search (Responses API / file_search)
rag:
  tool:
    - okp
okp:
  rhokp_url: http://my-okp-instance:8080
  offline: true
  search_mode: keyword

# rag.inline — air-gap with keyword search (streaming_query)
rag:
  inline:
    - okp
okp:
  rhokp_url: http://my-okp-instance:8080
  offline: true
  search_mode: keyword

Tradeoff: Keyword search returns less precise document references than hybrid/semantic (matches on term frequency, not meaning), but the LLM answer quality is still good. For air-gapped deployments, this is the acceptable tradeoff to avoid the HuggingFace dependency.

Type of change

• Refactor
• New feature
• Bug fix
• CVE fix
• Optimization
• Documentation Update
• Configuration Update
• Bump-up service version
• Bump-up dependent library [pyproject.toml + uv.lock]
• Bump-up dependent library [requirements.*.txt for Konflux]
• Bump-up library or tool used for development (does not change the final image)
• CI configuration change
• Konflux configuration change
• Unit tests improvement
• Integration tests improvement
• End to end tests improvement
• Benchmarks improvement

Tools used to create PR

Identify any AI code assistants used in this PR (for transparency and review context)

• Assisted-by: Claude Code (Claude Opus 4.6)
• Generated by: N/A

Related Tickets & Documents

• Related Issue: https://redhat.atlassian.net/browse/RHIDP-14130
• Related PR: RHIDP-16060: migrate Solr vector_io provider from llama-stack to ogx lightspeed-providers#160

Checklist before requesting a review

• I have performed a self-review of my code.
• PR has passed all pre-merge test jobs.
• If it is a core feature, I have added thorough tests.

How to test

1. Automated tests

uv sync --group dev --group llslibdev
uv run make test-unit       # 3157 tests, all passing
uv run make verify           # linters + type checks

2. Verify OpenAPI schema is up to date

uv run scripts/generate_openapi_schema.py
diff <(cat docs/devel_doc/openapi.json) <(cat /tmp/openapi.json)  # should be empty

3. Test search_mode config parsing (manual)

Add search_mode to any config that has an okp: section:

okp:
  rhokp_url: http://my-okp-instance:8080
  offline: true
  search_mode: keyword   # new field: vector (default), keyword, or hybrid

Start the service — it should boot without errors. Invalid values (e.g. search_mode: foo) should be rejected by Pydantic validation.

4. Test with a local OKP instance (optional, requires OKP/Solr)

If you have a running OKP instance, you can test all RAG + search_mode combinations:

rag.tool (default vector search via Responses API):

rag:
  tool:
    - okp
okp:
  rhokp_url: http://localhost:8080
  offline: true

Ask a question via the Responses API — the file_search tool should return OKP chunks with reference_url links. The embedding model will be downloaded.

rag.tool + keyword (air-gap simulation via Responses API):

rag:
  tool:
    - okp
okp:
  rhokp_url: http://localhost:8080
  offline: true
  search_mode: keyword

Ask a question via the Responses API — file_search should return OKP chunks using BM25 keyword search. No embedding model should be downloaded (check ~/.cache/huggingface/hub/ is empty).

rag.inline + keyword (air-gap simulation via streaming_query):

rag:
  inline:
    - okp
okp:
  rhokp_url: http://localhost:8080
  offline: true
  search_mode: keyword

Ask a question via streaming_query — OKP results should come back using BM25 keyword search. No embedding model should be downloaded.

5. What was tested locally and on OpenShift

Local testing (all 4 scenarios):

Test	Config	Result
rag.tool (default vector)	rag.tool: [okp]	10 OKP sources, embedding model downloaded
rag.tool + keyword	rag.tool: [okp] + search_mode: keyword	10 OKP sources, no embedding model
rag.inline + keyword	rag.inline: [okp] + search_mode: keyword	3 OKP sources, no embedding model
rag.inline (default)	rag.inline: [okp]	OKP sources with hybrid search

OpenShift testing (RHDH + OKP on OpenShift cluster with Solr, unified synthesis mode):

Air-gap verification: with search_mode: keyword, the container's HuggingFace cache (/tmp/hf_cache/) does not exist — confirming no embedding model was downloaded. Without keyword mode, the cache contains the expected embedding model.

Test	Config	Result	Air-gap safe
rag.tool (default vector)	rag.tool: [okp]	10 OKP sources with valid docs.redhat.com URLs, embedding model downloaded	No
rag.tool + keyword	rag.tool: [okp] + search_mode: keyword	10 OKP sources with valid docs.redhat.com URLs, no embedding model downloaded	Yes
rag.inline + keyword	rag.inline: [okp] + search_mode: keyword	OKP sources returned via BM25 search, no embedding model downloaded	Yes

Summary by CodeRabbit

• New Features

  • Added configurable OKP search modes: semantic, hybrid, and keyword.
  • Added support for keyword search in Solr vector queries, including BM25 behavior.
  • Added automatic search-mode defaults and terminology mapping.
  • Improved referenced-document links for OKP and Solr results, including offline configurations.
  • Updated API documentation with the new search options and defaults.

• Tests

  • Expanded coverage for search modes, configuration behavior, and document URL generation.

[coderabbitai]

Walkthrough

The change updates Solr provider and embedding registration, adds OKP search-mode configuration and dispatch normalization, and constructs URLs for OKP referenced documents. OpenAPI schemas and unit tests reflect the new behavior.

Changes

OKP and Solr integration

Layer / File(s)	Summary
Solr provider and embedding registration deploy/lightspeed-stack/Containerfile, providers, src/constants.py, src/llama_stack_configuration.py, tests/unit/test_llama_stack_configuration.py	Provider references are updated. Solr uses sentence-transformers/solr_embedding through a shared constant.
Search-mode contracts and dispatch docs/devel_doc/openapi.json, src/models/common/query.py, src/models/config.py, src/constants.py, src/utils/vector_search.py, src/llama_stack_configuration.py, tests/unit/models/config/test_dump_configuration.py, tests/unit/utils/test_vector_search.py, tests/unit/test_llama_stack_configuration.py	OKP configuration and Solr requests accept keyword. Dispatch applies configured-mode precedence and maps lexical to keyword.
OKP document URL resolution src/utils/responses.py, src/utils/agents/tool_processor.py, tests/unit/utils/test_responses.py, tests/unit/utils/agents/test_tool_processor.py	OKP URLs use online references, offline source paths, or document IDs. Non-OKP URL handling remains unchanged.

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

Sequence Diagram(s)

sequenceDiagram
  participant Request as SolrVectorSearchRequest
  participant VectorSearch as vector_search
  participant Configuration as OkpConfiguration
  participant LlamaStack as Llama Stack
  Request->>VectorSearch: provide optional search mode
  VectorSearch->>Configuration: read okp.search_mode when absent
  VectorSearch->>LlamaStack: send normalized search mode
  Configuration->>LlamaStack: register default_search_mode

Loading

sequenceDiagram
  participant RAGResult
  participant ResponseParser as parse_referenced_documents
  participant URLBuilder as _build_okp_doc_url
  participant ToolProcessor as build_referenced_document
  participant ReferencedDocument
  RAGResult->>ResponseParser: provide chunk metadata
  ResponseParser->>URLBuilder: build OKP document URL
  RAGResult->>ToolProcessor: provide result
  ToolProcessor->>URLBuilder: build OKP URL when needed
  ToolProcessor->>ReferencedDocument: set resolved source and URL

Loading

Suggested reviewers: are-ces, anik120, asimurka

🚥 Pre-merge checks | ✅ 7

✅ Passed checks (7 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly summarizes the main changes: OKP citation URLs, configurable search modes, and the OGX provider migration.
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.
Performance And Algorithmic Complexity	✅ Passed	The changed paths add only constant-time mode mapping and per-result URL construction; no new API calls, list-then-query pattern, unbounded cache, or missing pagination/limit was introduced.
Security And Secret Handling	✅ Passed	No security violations found: no credential literals or Secret manifests were added; affected APIs retain auth/authz; Solr filters validate fields and escape values; URL code only builds citation l...

✨ Finishing Touches 💡 1

🛠️ Fix failing CI checks 💡

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests

✨ Simplify code

• Create PR with simplified code

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[coderabbitai]

Actionable comments posted: 3

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

deploy/lightspeed-stack/Containerfile (1)

39-48: 🗄️ Data Integrity & Integration | 🟠 Major | ⚡ Quick win

Align the Konflux artifact lock with LIGHTSPEED_PROVIDERS_COMMIT.

.konflux/artifacts.lock.yaml pins lightspeed-providers.zip to 8cd1b3d3bdd841ea99d31b334ae00a275581661c, while the Containerfile fallback uses faf6a89a3ad7856e2e7a934324f31d146108acdb. When cachi2 supplies the prefetched zip, extraction will not use the new commit, defeating the pin. Update the Konflux artifacts lock to the same commit before relying on this path.

🤖 Prompt for 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.

In `@deploy/lightspeed-stack/Containerfile` around lines 39 - 48, Update the
Konflux artifacts lock entry for lightspeed-providers.zip to pin commit
faf6a89a3ad7856e2e7a934324f31d146108acdb, matching the
LIGHTSPEED_PROVIDERS_COMMIT value used by the Containerfile fallback.

🤖 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 `@src/models/common/query.py`:
- Around line 147-154: Update the mode field description to document that
_build_query_params first uses configuration.okp.search_mode when mode is unset,
falling back to SOLR_VECTOR_SEARCH_DEFAULT_MODE only when that configuration is
also unset. Also update the deprecated-payload warning near the mode validation
logic to include "keyword" alongside the existing deprecated modes.

In `@src/utils/agents/tool_processor.py`:
- Around line 297-299: Update the OKP branch in the document URL processing flow
around _build_okp_doc_url so its result replaces the existing doc_url only when
a non-None URL is returned; otherwise preserve the standard URL already resolved
earlier. Add a regression test for an OKP result containing doc_url but no
source_path, reference_url, or doc_id, and assert the original URL remains.

In `@src/utils/responses.py`:
- Around line 861-865: In the `_build_okp_doc_url` docstring, replace the
`Args:` parameter header with the project-standard `Parameters:` header while
preserving the existing parameter description and return documentation.

---

Outside diff comments:
In `@deploy/lightspeed-stack/Containerfile`:
- Around line 39-48: Update the Konflux artifacts lock entry for
lightspeed-providers.zip to pin commit faf6a89a3ad7856e2e7a934324f31d146108acdb,
matching the LIGHTSPEED_PROVIDERS_COMMIT value used by the Containerfile
fallback.

🪄 Autofix

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: Pro Plus

Run ID: f370ac49-89c7-4d7a-b82e-40326b825161

📥 Commits

Reviewing files that changed from the base of the PR and between db92c22 and 88bc6c9.

📒 Files selected for processing (15)

• deploy/lightspeed-stack/Containerfile
• docs/devel_doc/openapi.json
• providers
• src/constants.py
• src/llama_stack_configuration.py
• src/models/common/query.py
• src/models/config.py
• src/utils/agents/tool_processor.py
• src/utils/responses.py
• src/utils/vector_search.py
• tests/unit/models/config/test_dump_configuration.py
• tests/unit/test_llama_stack_configuration.py
• tests/unit/utils/agents/test_tool_processor.py
• tests/unit/utils/test_responses.py
• tests/unit/utils/test_vector_search.py

📜 Review details

⏰ Context from checks skipped due to timeout. (14)

• GitHub Check: Pylinter
• GitHub Check: build-pr
• GitHub Check: integration_tests (3.13)
• GitHub Check: integration_tests (3.12)
• GitHub Check: unit_tests (3.13)
• GitHub Check: unit_tests (3.12)
• GitHub Check: E2E: library mode / ci / group 2
• GitHub Check: E2E Tests for Lightspeed Evaluation job
• GitHub Check: E2E: server mode / ci / group 3
• GitHub Check: E2E: library mode / ci / group 1
• GitHub Check: E2E: server mode / ci / group 2
• GitHub Check: E2E: library mode / ci / group 3
• GitHub Check: E2E: server mode / ci / group 1
• GitHub Check: Konflux kflux-prd-rh02 / lightspeed-stack-0-7-on-pull-request

⚠️ CI failures not shown inline (1)

GitHub Check: Konflux kflux-prd-rh02 / lightspeed-stack-0-8-on-pull-request: Failed

Conclusion: failure

View job details

Konflux kflux-prd-rh02/lightspeed-stack-0-8-on-pull-request has <b>failed</b>.
<ul>
<li><b>Namespace</b>: <a href="https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant">lightspeed-core-tenant</a></li>
<li><b>PipelineRun:</b> <a href="https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-vrwql">lightspeed-stack-0-8-on-pull-request-vrwql</a></li>
</ul>
<hr>
<h4>Task Statuses:</h4>
<table>
  <tr><th>Status</th><th>Duration</th><th>Name</th></tr>
<tr>
<td>🟢 Succeeded</td>
<td>4 seconds</td><td>
[init](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-vrwql/logs/init)
</td></tr>
<tr>
<td>🟢 Succeeded</td>
<td>14 seconds</td><td>
[clone-repository](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-vrwql/logs/clone-repository)
</td></tr>
<tr>
<td>🟢 Succeeded</td>
<td>11 seconds</td><td>
[prefetch-dependencies](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-vrwql/logs/prefetch-dependencies)
</td></tr>
<tr>
<td>🔴 Failed</td>
<td>10 seconds</td><td>
[build-images](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-vrwql/logs/build-images)
</td></tr>
</table>

🧰 Additional context used

📓 Path-based instructions (4)

**/*

📄 CodeRabbit inference engine (Custom checks)

**/*: Flag meaningful O(n^2)+ algorithms on non-trivial inputs, including handlers and Kubernetes list operations.
Flag N+1 patterns that list items and then query once per item, including Kubernetes API and database access.
Flag expensive work inside loops, including API calls, JSON parsing, and regex compilation.
Flag unbounded growth in caches, watchers, or buffers when eviction or limits are missing.
Flag missing pagination or limits on list operations and API endpoints.
Flag secrets or tokens logged in plaintext or hardcoded in source.
Flag API endpoints missing authentication or authorization.
Flag injection vulnerabilities, including SQL injection, command injection, and path traversal.
Flag sensitive data leaked in API responses, WebSocket messages, or logs.
Flag Kubernetes Secrets and Red Hat secrets missing OwnerReferences.

**/*: Use uv run for project commands; use uv sync --group dev --group llslibdev to install development dependencies.
Before completion, run uv run make format, uv run make verify, create unit tests for new code, and ensure tests pass.
Never commit secrets or keys; use environment variables for sensitive data.

Files:

• providers
• deploy/lightspeed-stack/Containerfile
• tests/unit/test_llama_stack_configuration.py
• src/utils/responses.py
• src/utils/vector_search.py
• src/llama_stack_configuration.py
• src/models/config.py
• src/utils/agents/tool_processor.py
• tests/unit/utils/agents/test_tool_processor.py
• src/models/common/query.py
• tests/unit/models/config/test_dump_configuration.py
• docs/devel_doc/openapi.json
• src/constants.py
• tests/unit/utils/test_vector_search.py
• tests/unit/utils/test_responses.py

tests/unit/**/*.py

📄 CodeRabbit inference engine (AGENTS.md)

tests/unit/**/*.py: Use pytest for all unit tests; do not use unittest.
Use conftest.py for shared pytest fixtures and pytest-mock for AsyncMock objects.
Mark asynchronous pytest tests with pytest.mark.asyncio.

Files:

• tests/unit/test_llama_stack_configuration.py
• tests/unit/utils/agents/test_tool_processor.py
• tests/unit/models/config/test_dump_configuration.py
• tests/unit/utils/test_vector_search.py
• tests/unit/utils/test_responses.py

**/*.{py,pyi}

📄 CodeRabbit inference engine (AGENTS.md)

Run and satisfy the configured Black, Pylint, Pyright, Ruff, pydocstyle, and type-checking standards.

Files:

• tests/unit/test_llama_stack_configuration.py
• src/utils/responses.py
• src/utils/vector_search.py
• src/llama_stack_configuration.py
• src/models/config.py
• src/utils/agents/tool_processor.py
• tests/unit/utils/agents/test_tool_processor.py
• src/models/common/query.py
• tests/unit/models/config/test_dump_configuration.py
• src/constants.py
• tests/unit/utils/test_vector_search.py
• tests/unit/utils/test_responses.py

src/**/*.py

📄 CodeRabbit inference engine (AGENTS.md)

src/**/*.py: Use absolute imports for internal Python modules.
All modules must start with descriptive docstrings; use logger = get_logger(__name__) from log.py for module logging.
Define shared constants in the central constants.py module, check it before adding constants, and annotate constants with Final[type].
Define type aliases at module level for clarity.
Use complete type annotations for function parameters and return types; prefer modern union syntax (str | int) and Optional[Type] for optional values.
Use typing_extensions.Self for model validators.
Use descriptive, action-oriented snake_case function names, such as get_, validate_, and check_.
Functions must not modify mutable parameters in place; return a new data structure instead.
Use async def for I/O operations and external API calls.
Handle APIConnectionError from Llama Stack.
Use from log import get_logger and initialize module logging with logger = get_logger(__name__).
Use standard log levels appropriately: debug for diagnostics, info for general execution, warning for unexpected or potential problems, and error for serious failures.
All classes must have descriptive docstrings, use PascalCase names, and use standard suffixes such as Configuration, Error/Exception, Resolver, and Interface where applicable.
Pydantic configuration models must extend ConfigurationBase; data models must extend BaseModel.
Abstract interfaces must use ABC with @abstractmethod decorators.
Use @model_validator and @field_validator for Pydantic model validation.
Class attributes must have complete, specific type annotations; avoid Any.
Follow Google Python docstring conventions; provide required docstrings for modules, classes, and functions, including relevant Parameters, Returns, Raises, and Attributes sections.

Files:

• src/utils/responses.py
• src/utils/vector_search.py
• src/llama_stack_configuration.py
• src/models/config.py
• src/utils/agents/tool_processor.py
• src/models/common/query.py
• src/constants.py

🧠 Learnings (6)

📚 Learning: 2026-06-24T13:45:37.249Z

Learnt from: Jdubrick
Repo: lightspeed-core/lightspeed-stack PR: 1971
File: src/utils/markdown_repair.py:31-36
Timestamp: 2026-06-24T13:45:37.249Z
Learning: In the lightspeed-stack repository, docstrings must use the section header name "Parameters:" (not "Args:") for function arguments, even if the project references Google Python docstring conventions. Ensure docstrings follow the project’s established "Parameters:" header format for any documented function parameters.

Applied to files:

• tests/unit/test_llama_stack_configuration.py
• src/utils/responses.py
• src/utils/vector_search.py
• src/llama_stack_configuration.py
• src/models/config.py
• src/utils/agents/tool_processor.py
• tests/unit/utils/agents/test_tool_processor.py
• src/models/common/query.py
• tests/unit/models/config/test_dump_configuration.py
• src/constants.py
• tests/unit/utils/test_vector_search.py
• tests/unit/utils/test_responses.py

📚 Learning: 2026-02-23T14:56:59.186Z

Learnt from: asimurka
Repo: lightspeed-core/lightspeed-stack PR: 1198
File: src/utils/responses.py:184-192
Timestamp: 2026-02-23T14:56:59.186Z
Learning: In the lightspeed-stack codebase (lightspeed-core/lightspeed-stack), do not enforce de-duplication of duplicate client.models.list() calls in model selection flows (e.g., in src/utils/responses.py prepare_responses_params). These calls are considered relatively cheap and removing duplicates could add unnecessary complexity to the flow. Apply this guideline specifically to this file/context unless similar performance characteristics and design decisions are documented elsewhere.

Applied to files:

• src/utils/responses.py

📚 Learning: 2026-07-06T15:26:18.398Z

Learnt from: Jdubrick
Repo: lightspeed-core/lightspeed-stack PR: 2071
File: src/models/config.py:2416-2422
Timestamp: 2026-07-06T15:26:18.398Z
Learning: In this repo’s Python code under src/**, don’t treat differences in string concatenation style as a style inconsistency when Black has effectively forced (or made clearer) use of explicit `+` string concatenation in multi-line logger/string expressions. If adjacent-literal implicit concatenation is avoided/changed specifically to accommodate Black’s formatting in these call sites, accept the `+` usage and don’t recommend converting it solely for consistency with nearby blocks that use implicit concatenation.

Applied to files:

• src/utils/responses.py
• src/utils/vector_search.py
• src/llama_stack_configuration.py
• src/models/config.py
• src/utils/agents/tool_processor.py
• src/models/common/query.py
• src/constants.py

📚 Learning: 2026-07-17T19:25:05.325Z

Learnt from: Jdubrick
Repo: lightspeed-core/lightspeed-stack PR: 2166
File: src/utils/saved_prompts.py:129-157
Timestamp: 2026-07-17T19:25:05.325Z
Learning: For any endpoint that handles saved prompts and calls `src/utils/saved_prompts.py::create_saved_prompt`, treat the endpoint as the validation boundary. Before calling `create_saved_prompt`, validate the incoming saved-prompt name and content, specifically using `validate_saved_prompt_name` and then persist (store) the normalized value it returns. Do not call `create_saved_prompt` with unvalidated/raw name/content.

Applied to files:

• src/utils/responses.py
• src/utils/vector_search.py
• src/llama_stack_configuration.py
• src/models/config.py
• src/utils/agents/tool_processor.py
• src/models/common/query.py
• src/constants.py

📚 Learning: 2026-01-12T10:58:40.230Z

Learnt from: blublinsky
Repo: lightspeed-core/lightspeed-stack PR: 972
File: src/models/config.py:459-513
Timestamp: 2026-01-12T10:58:40.230Z
Learning: In lightspeed-core/lightspeed-stack, for Python files under src/models, when a user claims a fix is done but the issue persists, verify the current code state before accepting the fix. Steps: review the diff, fetch the latest changes, run relevant tests, reproduce the issue, search the codebase for lingering references to the original problem, confirm the fix is applied and not undone by subsequent commits, and validate with local checks to ensure the issue is resolved.

Applied to files:

• src/models/config.py
• src/models/common/query.py

📚 Learning: 2026-02-25T07:46:33.545Z

Learnt from: asimurka
Repo: lightspeed-core/lightspeed-stack PR: 1211
File: src/models/responses.py:8-16
Timestamp: 2026-02-25T07:46:33.545Z
Learning: In the Python codebase, requests.py should use OpenAIResponseInputTool as Tool while responses.py uses OpenAIResponseTool as Tool. This difference is intentional due to differing schemas for input vs output tools in llama-stack-api. Apply this distinction consistently to other models under src/models (e.g., ensure request-related tools use the InputTool variant and response-related tools use the ResponseTool variant). If adding new tools, choose the corresponding InputTool or Tool class based on whether the tool represents input or output, and document the rationale in code comments.

Applied to files:

• src/models/config.py
• src/models/common/query.py

🔇 Additional comments (13)

providers (1)

1-1: 🗄️ Data Integrity & Integration

No change needed.

The providers submodule pointer resolves to the expected commit, and the commit is available from the configured remote.

src/constants.py (1)

238-239: LGTM!

Also applies to: 257-257

src/llama_stack_configuration.py (1)

885-885: LGTM!

Also applies to: 911-911

tests/unit/test_llama_stack_configuration.py (1)

811-811: LGTM!

docs/devel_doc/openapi.json (1)

15953-15969: LGTM!

Also applies to: 21297-21310

src/models/config.py (1)

2558-2566: LGTM!

src/utils/vector_search.py (1)

124-128: LGTM!

tests/unit/utils/test_vector_search.py (1)

165-174: LGTM!

Also applies to: 198-250

tests/unit/models/config/test_dump_configuration.py (1)

251-251: LGTM!

Also applies to: 481-481, 862-862, 1127-1127, 1425-1425, 1650-1650, 2035-2035, 2266-2266, 2497-2497, 2735-2735

src/utils/responses.py (1)

8-8: LGTM!

Also applies to: 867-878, 930-935

src/utils/agents/tool_processor.py (1)

20-20: LGTM!

Also applies to: 32-32, 290-295, 310-310

tests/unit/utils/test_responses.py (1)

72-72: LGTM!

Also applies to: 3150-3378

tests/unit/utils/agents/test_tool_processor.py (1)

346-404: LGTM!

[tisnik]

LGTM

[coderabbitai]

Actionable comments posted: 1

🤖 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 `@src/models/common/query.py`:
- Around line 147-153: Update the class docstring for the query mode near the
mode field to state that an omitted mode first uses
configuration.okp.search_mode, then falls back to
SOLR_VECTOR_SEARCH_DEFAULT_MODE. Keep the existing descriptions of explicit
modes and search behavior unchanged, and ensure the class-level documentation
matches the field description.

🪄 Autofix

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: Pro Plus

Run ID: 3ce675ff-ac87-4402-b5c6-d0cd1f54b420

📥 Commits

Reviewing files that changed from the base of the PR and between 88bc6c9 and 223fb7f.

📒 Files selected for processing (4)

• docs/devel_doc/openapi.json
• src/models/common/query.py
• src/utils/agents/tool_processor.py
• src/utils/responses.py

📜 Review details

⏰ Context from checks skipped due to timeout. (10)

• GitHub Check: Pylinter
• GitHub Check: E2E: library mode / ci / group 1
• GitHub Check: E2E: server mode / ci / group 3
• GitHub Check: E2E Tests for Lightspeed Evaluation job
• GitHub Check: unit_tests (3.12)
• GitHub Check: E2E: server mode / ci / group 1
• GitHub Check: E2E: server mode / ci / group 2
• GitHub Check: unit_tests (3.13)
• GitHub Check: check
• GitHub Check: Konflux kflux-prd-rh02 / lightspeed-stack-0-7-on-pull-request

⚠️ CI failures not shown inline (1)

GitHub Check: Konflux kflux-prd-rh02 / lightspeed-stack-0-8-on-pull-request: Failed

Conclusion: failure

View job details

Konflux kflux-prd-rh02/lightspeed-stack-0-8-on-pull-request has <b>failed</b>.
<ul>
<li><b>Namespace</b>: <a href="https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant">lightspeed-core-tenant</a></li>
<li><b>PipelineRun:</b> <a href="https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-4rjp9">lightspeed-stack-0-8-on-pull-request-4rjp9</a></li>
</ul>
<hr>
<h4>Task Statuses:</h4>
<table>
  <tr><th>Status</th><th>Duration</th><th>Name</th></tr>
<tr>
<td>🟢 Succeeded</td>
<td>5 seconds</td><td>
[init](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-4rjp9/logs/init)
</td></tr>
<tr>
<td>🟢 Succeeded</td>
<td>14 seconds</td><td>
[clone-repository](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-4rjp9/logs/clone-repository)
</td></tr>
<tr>
<td>🟢 Succeeded</td>
<td>9 seconds</td><td>
[prefetch-dependencies](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-4rjp9/logs/prefetch-dependencies)
</td></tr>
<tr>
<td>🔴 Failed</td>
<td>11 seconds</td><td>
[build-images](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-4rjp9/logs/build-images)
</td></tr>
</table>

🧰 Additional context used

📓 Path-based instructions (5)

**/*

📄 CodeRabbit inference engine (Custom checks)

**/*: Flag meaningful O(n^2)+ algorithms on non-trivial inputs, including handlers and Kubernetes list operations.
Flag N+1 patterns that list items and then query once per item, including Kubernetes API and database access.
Flag expensive work inside loops, including API calls, JSON parsing, and regex compilation.
Flag unbounded growth in caches, watchers, or buffers when eviction or limits are missing.
Flag missing pagination or limits on list operations and API endpoints.
Flag secrets or tokens logged in plaintext or hardcoded in source.
Flag API endpoints missing authentication or authorization.
Flag injection vulnerabilities, including SQL injection, command injection, and path traversal.
Flag sensitive data leaked in API responses, WebSocket messages, or logs.
Flag Kubernetes Secrets and Red Hat secrets missing OwnerReferences.

**/*: Use uv for dependency management and prefix project commands with uv run; synchronize development dependencies with uv sync --group dev --group llslibdev when required.
Before completing code changes, run uv run make format, uv run make verify, and the applicable tests; regenerate the OpenAPI schema when models change, and report completion only after checks pass.
Follow existing code patterns in the module being modified and write unit tests covering new functionality.

Files:

• src/utils/agents/tool_processor.py
• src/utils/responses.py
• src/models/common/query.py
• docs/devel_doc/openapi.json

**/*.py

📄 CodeRabbit inference engine (AGENTS.md)

**/*.py: Use absolute imports for internal Python modules.
All modules must begin with descriptive docstrings; package __init__.py files must contain brief package descriptions.
Use logger = get_logger(__name__) from log.py for module logging.
Use Final[type] type hints for constants and define shared constants in the central constants.py module.

Files:

• src/utils/agents/tool_processor.py
• src/utils/responses.py
• src/models/common/query.py

src/**/*.py

📄 CodeRabbit inference engine (AGENTS.md)

src/**/*.py: Functions must have descriptive docstrings, complete parameter and return type annotations, and descriptive snake_case action-oriented names.
Use typing_extensions.Self for model validators, modern union syntax such as str | int, and Optional[Type] for optional types.
Avoid modifying mutable parameters in place; return a new data structure instead.
Use async def for I/O operations and external API calls.
Classes require descriptive docstrings, PascalCase descriptive names, complete attribute type annotations, and specific types instead of Any. Use ABC and @abstractmethod for interfaces.
Follow Google Python docstring conventions and include relevant Parameters, Returns, Raises, and Attributes sections.

Files:

• src/utils/agents/tool_processor.py
• src/utils/responses.py
• src/models/common/query.py

**/*.{py,yaml,yml,json,toml}

📄 CodeRabbit inference engine (AGENTS.md)

Never commit secrets or keys; use environment variables for sensitive data.

Files:

• src/utils/agents/tool_processor.py
• src/utils/responses.py
• src/models/common/query.py
• docs/devel_doc/openapi.json

src/models/**/*.py

📄 CodeRabbit inference engine (AGENTS.md)

Pydantic configuration models must extend ConfigurationBase; data models must extend BaseModel; use model_validator and field_validator for validation.

Files:

• src/models/common/query.py

🧠 Learnings (6)

📚 Learning: 2026-06-24T13:45:37.249Z

Learnt from: Jdubrick
Repo: lightspeed-core/lightspeed-stack PR: 1971
File: src/utils/markdown_repair.py:31-36
Timestamp: 2026-06-24T13:45:37.249Z
Learning: In the lightspeed-stack repository, docstrings must use the section header name "Parameters:" (not "Args:") for function arguments, even if the project references Google Python docstring conventions. Ensure docstrings follow the project’s established "Parameters:" header format for any documented function parameters.

Applied to files:

• src/utils/agents/tool_processor.py
• src/utils/responses.py
• src/models/common/query.py

📚 Learning: 2026-07-06T15:26:18.398Z

Learnt from: Jdubrick
Repo: lightspeed-core/lightspeed-stack PR: 2071
File: src/models/config.py:2416-2422
Timestamp: 2026-07-06T15:26:18.398Z
Learning: In this repo’s Python code under src/**, don’t treat differences in string concatenation style as a style inconsistency when Black has effectively forced (or made clearer) use of explicit `+` string concatenation in multi-line logger/string expressions. If adjacent-literal implicit concatenation is avoided/changed specifically to accommodate Black’s formatting in these call sites, accept the `+` usage and don’t recommend converting it solely for consistency with nearby blocks that use implicit concatenation.

Applied to files:

• src/utils/agents/tool_processor.py
• src/utils/responses.py
• src/models/common/query.py

📚 Learning: 2026-07-17T19:25:05.325Z

Learnt from: Jdubrick
Repo: lightspeed-core/lightspeed-stack PR: 2166
File: src/utils/saved_prompts.py:129-157
Timestamp: 2026-07-17T19:25:05.325Z
Learning: For any endpoint that handles saved prompts and calls `src/utils/saved_prompts.py::create_saved_prompt`, treat the endpoint as the validation boundary. Before calling `create_saved_prompt`, validate the incoming saved-prompt name and content, specifically using `validate_saved_prompt_name` and then persist (store) the normalized value it returns. Do not call `create_saved_prompt` with unvalidated/raw name/content.

Applied to files:

• src/utils/agents/tool_processor.py
• src/utils/responses.py
• src/models/common/query.py

📚 Learning: 2026-02-23T14:56:59.186Z

Learnt from: asimurka
Repo: lightspeed-core/lightspeed-stack PR: 1198
File: src/utils/responses.py:184-192
Timestamp: 2026-02-23T14:56:59.186Z
Learning: In the lightspeed-stack codebase (lightspeed-core/lightspeed-stack), do not enforce de-duplication of duplicate client.models.list() calls in model selection flows (e.g., in src/utils/responses.py prepare_responses_params). These calls are considered relatively cheap and removing duplicates could add unnecessary complexity to the flow. Apply this guideline specifically to this file/context unless similar performance characteristics and design decisions are documented elsewhere.

Applied to files:

• src/utils/responses.py

📚 Learning: 2026-01-12T10:58:40.230Z

Learnt from: blublinsky
Repo: lightspeed-core/lightspeed-stack PR: 972
File: src/models/config.py:459-513
Timestamp: 2026-01-12T10:58:40.230Z
Learning: In lightspeed-core/lightspeed-stack, for Python files under src/models, when a user claims a fix is done but the issue persists, verify the current code state before accepting the fix. Steps: review the diff, fetch the latest changes, run relevant tests, reproduce the issue, search the codebase for lingering references to the original problem, confirm the fix is applied and not undone by subsequent commits, and validate with local checks to ensure the issue is resolved.

Applied to files:

• src/models/common/query.py

📚 Learning: 2026-02-25T07:46:33.545Z

Learnt from: asimurka
Repo: lightspeed-core/lightspeed-stack PR: 1211
File: src/models/responses.py:8-16
Timestamp: 2026-02-25T07:46:33.545Z
Learning: In the Python codebase, requests.py should use OpenAIResponseInputTool as Tool while responses.py uses OpenAIResponseTool as Tool. This difference is intentional due to differing schemas for input vs output tools in llama-stack-api. Apply this distinction consistently to other models under src/models (e.g., ensure request-related tools use the InputTool variant and response-related tools use the ResponseTool variant). If adding new tools, choose the corresponding InputTool or Tool class based on whether the tool represents input or output, and document the rationale in code comments.

Applied to files:

• src/models/common/query.py

🔇 Additional comments (4)

docs/devel_doc/openapi.json (1)

21306-21306: LGTM!

src/models/common/query.py (1)

210-210: LGTM!

src/utils/responses.py (1)

8-8: LGTM!

Also applies to: 854-878, 930-935

src/utils/agents/tool_processor.py (1)

20-20: LGTM!

Also applies to: 32-32, 290-300, 310-310

[coderabbitai]

📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

Synchronize the class-level mode documentation.

The field description now documents the configured OKP fallback, but the class docstring at Lines 138 and 142 still says that an omitted mode uses the server default. src/utils/vector_search.py uses configuration.okp.search_mode before SOLR_VECTOR_SEARCH_DEFAULT_MODE. Update the class docstring to describe the same precedence.

🤖 Prompt for 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.

In `@src/models/common/query.py` around lines 147 - 153, Update the class
docstring for the query mode near the mode field to state that an omitted mode
first uses configuration.okp.search_mode, then falls back to
SOLR_VECTOR_SEARCH_DEFAULT_MODE. Keep the existing descriptions of explicit
modes and search behavior unchanged, and ensure the class-level documentation
matches the field description.

[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 `@src/llama_stack_configuration.py`:
- Line 770: Update the docstring for enrich_solr to rename its parameter section
header from “Args:” to “Parameters:”, leaving the documented parameter content
unchanged.
- Around line 932-937: Update the configuration-building logic around ls_config
to avoid mutating the caller-owned object: create a copied configuration with
copied vector_stores and chunk_retrieval_params, set default_search_mode on the
copy, and return it. Update the enclosing function’s callers and tests to
consume the returned configuration rather than relying on in-place changes.

🪄 Autofix

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: Pro Plus

Run ID: 03d5279a-51ed-4df0-9f44-1d78414614a9

📥 Commits

Reviewing files that changed from the base of the PR and between 223fb7f and be2344d.

📒 Files selected for processing (2)

• src/llama_stack_configuration.py
• tests/unit/test_llama_stack_configuration.py

📜 Review details

⏰ Context from checks skipped due to timeout. (1)

• GitHub Check: Konflux kflux-prd-rh02 / lightspeed-stack-0-7-on-pull-request

⚠️ CI failures not shown inline (1)

GitHub Check: Konflux kflux-prd-rh02 / lightspeed-stack-0-8-on-pull-request: Failed

Conclusion: failure

View job details

Konflux kflux-prd-rh02/lightspeed-stack-0-8-on-pull-request has <b>failed</b>.
<ul>
<li><b>Namespace</b>: <a href="https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant">lightspeed-core-tenant</a></li>
<li><b>PipelineRun:</b> <a href="https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-t6sj7">lightspeed-stack-0-8-on-pull-request-t6sj7</a></li>
</ul>
<hr>
<h4>Task Statuses:</h4>
<table>
  <tr><th>Status</th><th>Duration</th><th>Name</th></tr>
<tr>
<td>🟢 Succeeded</td>
<td>5 seconds</td><td>
[init](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-t6sj7/logs/init)
</td></tr>
<tr>
<td>🟢 Succeeded</td>
<td>16 seconds</td><td>
[clone-repository](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-t6sj7/logs/clone-repository)
</td></tr>
<tr>
<td>🟢 Succeeded</td>
<td>10 seconds</td><td>
[prefetch-dependencies](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-t6sj7/logs/prefetch-dependencies)
</td></tr>
<tr>
<td>🔴 Failed</td>
<td>12 seconds</td><td>
[build-images](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-t6sj7/logs/build-images)
</td></tr>
</table>

🧰 Additional context used

📓 Path-based instructions (5)

**/*

📄 CodeRabbit inference engine (Custom checks)

**/*: Flag meaningful O(n^2)+ algorithms on non-trivial inputs, including handlers and Kubernetes list operations.
Flag N+1 patterns that list items and then query once per item, including Kubernetes API and database access.
Flag expensive work inside loops, including API calls, JSON parsing, and regex compilation.
Flag unbounded growth in caches, watchers, or buffers when eviction or limits are missing.
Flag missing pagination or limits on list operations and API endpoints.
Flag secrets or tokens logged in plaintext or hardcoded in source.
Flag API endpoints missing authentication or authorization.
Flag injection vulnerabilities, including SQL injection, command injection, and path traversal.
Flag sensitive data leaked in API responses, WebSocket messages, or logs.
Flag Kubernetes Secrets and Red Hat secrets missing OwnerReferences.

**/*: Use uv for dependency management and prefix project commands with uv run; synchronize development dependencies with uv sync --group dev --group llslibdev when required.
Before completing code changes, run uv run make format, uv run make verify, and the applicable tests; regenerate the OpenAPI schema when models change, and report completion only after checks pass.
Follow existing code patterns in the module being modified and write unit tests covering new functionality.

Files:

• tests/unit/test_llama_stack_configuration.py
• src/llama_stack_configuration.py

**/*.py

📄 CodeRabbit inference engine (AGENTS.md)

**/*.py: Use absolute imports for internal Python modules.
All modules must begin with descriptive docstrings; package __init__.py files must contain brief package descriptions.
Use logger = get_logger(__name__) from log.py for module logging.
Use Final[type] type hints for constants and define shared constants in the central constants.py module.

Files:

• tests/unit/test_llama_stack_configuration.py
• src/llama_stack_configuration.py

tests/unit/**/*.py

📄 CodeRabbit inference engine (AGENTS.md)

Use pytest for unit tests; do not use unittest. Put shared fixtures in conftest.py, use pytest-mock for mocks, mark asynchronous tests with pytest.mark.asyncio, and maintain at least 60% unit-test coverage.

Files:

• tests/unit/test_llama_stack_configuration.py

**/*.{py,yaml,yml,json,toml}

📄 CodeRabbit inference engine (AGENTS.md)

Never commit secrets or keys; use environment variables for sensitive data.

Files:

• tests/unit/test_llama_stack_configuration.py
• src/llama_stack_configuration.py

src/**/*.py

📄 CodeRabbit inference engine (AGENTS.md)

src/**/*.py: Functions must have descriptive docstrings, complete parameter and return type annotations, and descriptive snake_case action-oriented names.
Use typing_extensions.Self for model validators, modern union syntax such as str | int, and Optional[Type] for optional types.
Avoid modifying mutable parameters in place; return a new data structure instead.
Use async def for I/O operations and external API calls.
Classes require descriptive docstrings, PascalCase descriptive names, complete attribute type annotations, and specific types instead of Any. Use ABC and @abstractmethod for interfaces.
Follow Google Python docstring conventions and include relevant Parameters, Returns, Raises, and Attributes sections.

Files:

• src/llama_stack_configuration.py

🧠 Learnings (3)

📚 Learning: 2026-06-24T13:45:37.249Z

Learnt from: Jdubrick
Repo: lightspeed-core/lightspeed-stack PR: 1971
File: src/utils/markdown_repair.py:31-36
Timestamp: 2026-06-24T13:45:37.249Z
Learning: In the lightspeed-stack repository, docstrings must use the section header name "Parameters:" (not "Args:") for function arguments, even if the project references Google Python docstring conventions. Ensure docstrings follow the project’s established "Parameters:" header format for any documented function parameters.

Applied to files:

• tests/unit/test_llama_stack_configuration.py
• src/llama_stack_configuration.py

📚 Learning: 2026-07-06T15:26:18.398Z

Learnt from: Jdubrick
Repo: lightspeed-core/lightspeed-stack PR: 2071
File: src/models/config.py:2416-2422
Timestamp: 2026-07-06T15:26:18.398Z
Learning: In this repo’s Python code under src/**, don’t treat differences in string concatenation style as a style inconsistency when Black has effectively forced (or made clearer) use of explicit `+` string concatenation in multi-line logger/string expressions. If adjacent-literal implicit concatenation is avoided/changed specifically to accommodate Black’s formatting in these call sites, accept the `+` usage and don’t recommend converting it solely for consistency with nearby blocks that use implicit concatenation.

Applied to files:

• src/llama_stack_configuration.py

📚 Learning: 2026-07-17T19:25:05.325Z

Learnt from: Jdubrick
Repo: lightspeed-core/lightspeed-stack PR: 2166
File: src/utils/saved_prompts.py:129-157
Timestamp: 2026-07-17T19:25:05.325Z
Learning: For any endpoint that handles saved prompts and calls `src/utils/saved_prompts.py::create_saved_prompt`, treat the endpoint as the validation boundary. Before calling `create_saved_prompt`, validate the incoming saved-prompt name and content, specifically using `validate_saved_prompt_name` and then persist (store) the normalized value it returns. Do not call `create_saved_prompt` with unvalidated/raw name/content.

Applied to files:

• src/llama_stack_configuration.py

🔇 Additional comments (3)

src/llama_stack_configuration.py (2)

881-885: LGTM!

Also applies to: 911-911

---

922-931: LGTM!

tests/unit/test_llama_stack_configuration.py (1)

887-950: LGTM!

[coderabbitai]

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

src/llama_stack_configuration.py (1)

881-885: 🗄️ Data Integrity & Integration | 🟠 Major | 🏗️ Heavy lift

Normalize existing Solr registrations before returning enrich_solr.

enrich_solr appends the Solr vector store only when portal-rag is missing, so existing records can keep legacy embedding_model values. Its model duplicate check compares the stripped baseline model, so an existing model can block registration of sentence-transformers/solr_embedding. Add migration/upsert logic and cover it with an upgrade test.

🤖 Prompt for 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.

In `@src/llama_stack_configuration.py` around lines 881 - 885, Update enrich_solr
to normalize existing Solr vector-store registrations before returning: migrate
matching records to constants.SOLR_EMBEDDING_MODEL_ID and upsert the required
Solr registration instead of only appending when portal-rag is absent. Preserve
the duplicate-model check using the normalized model, and add an upgrade test
covering legacy embedding_model values and the resulting registration.

♻️ Duplicate comments (1)

src/llama_stack_configuration.py (1)

932-937: 🗄️ Data Integrity & Integration | 🟠 Major | 🏗️ Heavy lift

Stop mutating the caller-owned configuration.

setdefault and the direct assignment modify ls_config in place. This branch adds persistent state to the caller's dictionary and violates the repository rule for mutable parameters. Build an independent configuration, return it, and update callers and tests to consume the returned value.

🤖 Prompt for 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.

In `@src/llama_stack_configuration.py` around lines 932 - 937, Update the
configuration handling around chunk_retrieval_params so it does not mutate the
caller-owned ls_config: create an independent configuration before applying
vector_stores and default_search_mode changes, return that configuration, and
update all callers and tests to consume the returned value.

Source: Coding guidelines

🤖 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.

Outside diff comments:
In `@src/llama_stack_configuration.py`:
- Around line 881-885: Update enrich_solr to normalize existing Solr
vector-store registrations before returning: migrate matching records to
constants.SOLR_EMBEDDING_MODEL_ID and upsert the required Solr registration
instead of only appending when portal-rag is absent. Preserve the
duplicate-model check using the normalized model, and add an upgrade test
covering legacy embedding_model values and the resulting registration.

---

Duplicate comments:
In `@src/llama_stack_configuration.py`:
- Around line 932-937: Update the configuration handling around
chunk_retrieval_params so it does not mutate the caller-owned ls_config: create
an independent configuration before applying vector_stores and
default_search_mode changes, return that configuration, and update all callers
and tests to consume the returned value.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 17ca58fd-4177-4747-b7d5-ac6c037cb5fb

📥 Commits

Reviewing files that changed from the base of the PR and between be2344d and 8a6d9bb.

📒 Files selected for processing (2)

• src/llama_stack_configuration.py
• src/models/common/query.py

📜 Review details

⏰ Context from checks skipped due to timeout. (1)

• GitHub Check: Konflux kflux-prd-rh02 / lightspeed-stack-0-7-on-pull-request

⚠️ CI failures not shown inline (1)

GitHub Check: Konflux kflux-prd-rh02 / lightspeed-stack-0-8-on-pull-request: Failed

Conclusion: failure

View job details

Konflux kflux-prd-rh02/lightspeed-stack-0-8-on-pull-request has <b>failed</b>.
<ul>
<li><b>Namespace</b>: <a href="https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant">lightspeed-core-tenant</a></li>
<li><b>PipelineRun:</b> <a href="https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-z9zv4">lightspeed-stack-0-8-on-pull-request-z9zv4</a></li>
</ul>
<hr>
<h4>Task Statuses:</h4>
<table>
  <tr><th>Status</th><th>Duration</th><th>Name</th></tr>
<tr>
<td>🟢 Succeeded</td>
<td>5 seconds</td><td>
[init](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-z9zv4/logs/init)
</td></tr>
<tr>
<td>🟢 Succeeded</td>
<td>14 seconds</td><td>
[clone-repository](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-z9zv4/logs/clone-repository)
</td></tr>
<tr>
<td>🟢 Succeeded</td>
<td>10 seconds</td><td>
[prefetch-dependencies](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-z9zv4/logs/prefetch-dependencies)
</td></tr>
<tr>
<td>🔴 Failed</td>
<td>10 seconds</td><td>
[build-images](https://konflux-ui.apps.kflux-prd-rh02.0fk9.p1.openshiftapps.com/ns/lightspeed-core-tenant/pipelinerun/lightspeed-stack-0-8-on-pull-request-z9zv4/logs/build-images)
</td></tr>
</table>

🧰 Additional context used

📓 Path-based instructions (3)

**/*

📄 CodeRabbit inference engine (Custom checks)

**/*: Flag meaningful O(n^2)+ algorithms on non-trivial inputs, including handlers and Kubernetes list operations.
Flag N+1 patterns that list items and then query once per item, including Kubernetes API and database access.
Flag expensive work inside loops, including API calls, JSON parsing, and regex compilation.
Flag unbounded growth in caches, watchers, or buffers when eviction or limits are missing.
Flag missing pagination or limits on list operations and API endpoints.
Flag secrets or tokens logged in plaintext or hardcoded in source.
Flag API endpoints missing authentication or authorization.
Flag injection vulnerabilities, including SQL injection, command injection, and path traversal.
Flag sensitive data leaked in API responses, WebSocket messages, or logs.
Flag Kubernetes Secrets and Red Hat secrets missing OwnerReferences.

Never commit secrets or keys; use environment variables for sensitive data.

Files:

• src/models/common/query.py
• src/llama_stack_configuration.py

**/*.py

📄 CodeRabbit inference engine (AGENTS.md)

**/*.py: Use absolute imports for internal Python modules.
Every module must begin with a descriptive docstring explaining its purpose.
Use logger = get_logger(__name__) from log.py for module logging.
Define shared constants in the central constants.py module and annotate constants with Final[type].
All functions must have complete parameter and return type annotations and descriptive docstrings.
Use snake_case, descriptive, action-oriented names for functions, such as get_, validate_, and check_.
Avoid modifying mutable parameters in place; return a new data structure instead.
Use async def for I/O operations and external API calls.
Handle APIConnectionError from Llama Stack integrations.
Use standard logger levels appropriately: debug for diagnostics, info for general execution, warning for unexpected or potentially problematic conditions, and error for serious failures.
All classes must have descriptive docstrings, use PascalCase names, and provide complete, specific type annotations for class attributes.
Use Google Python docstring conventions, including Parameters, Returns, Raises, and Attributes sections when applicable.

Files:

• src/models/common/query.py
• src/llama_stack_configuration.py

src/models/**/*.py

📄 CodeRabbit inference engine (AGENTS.md)

src/models/**/*.py: Pydantic configuration models must extend ConfigurationBase; data models must extend BaseModel.
Use @field_validator and @model_validator for Pydantic model validation; model validators should use typing_extensions.Self where applicable.

Files:

• src/models/common/query.py

🧠 Learnings (5)

📚 Learning: 2026-01-12T10:58:40.230Z

Learnt from: blublinsky
Repo: lightspeed-core/lightspeed-stack PR: 972
File: src/models/config.py:459-513
Timestamp: 2026-01-12T10:58:40.230Z
Learning: In lightspeed-core/lightspeed-stack, for Python files under src/models, when a user claims a fix is done but the issue persists, verify the current code state before accepting the fix. Steps: review the diff, fetch the latest changes, run relevant tests, reproduce the issue, search the codebase for lingering references to the original problem, confirm the fix is applied and not undone by subsequent commits, and validate with local checks to ensure the issue is resolved.

Applied to files:

• src/models/common/query.py

📚 Learning: 2026-02-25T07:46:33.545Z

Learnt from: asimurka
Repo: lightspeed-core/lightspeed-stack PR: 1211
File: src/models/responses.py:8-16
Timestamp: 2026-02-25T07:46:33.545Z
Learning: In the Python codebase, requests.py should use OpenAIResponseInputTool as Tool while responses.py uses OpenAIResponseTool as Tool. This difference is intentional due to differing schemas for input vs output tools in llama-stack-api. Apply this distinction consistently to other models under src/models (e.g., ensure request-related tools use the InputTool variant and response-related tools use the ResponseTool variant). If adding new tools, choose the corresponding InputTool or Tool class based on whether the tool represents input or output, and document the rationale in code comments.

Applied to files:

• src/models/common/query.py

📚 Learning: 2026-06-24T13:45:37.249Z

Learnt from: Jdubrick
Repo: lightspeed-core/lightspeed-stack PR: 1971
File: src/utils/markdown_repair.py:31-36
Timestamp: 2026-06-24T13:45:37.249Z
Learning: In the lightspeed-stack repository, docstrings must use the section header name "Parameters:" (not "Args:") for function arguments, even if the project references Google Python docstring conventions. Ensure docstrings follow the project’s established "Parameters:" header format for any documented function parameters.

Applied to files:

• src/models/common/query.py
• src/llama_stack_configuration.py

📚 Learning: 2026-07-06T15:26:18.398Z

Learnt from: Jdubrick
Repo: lightspeed-core/lightspeed-stack PR: 2071
File: src/models/config.py:2416-2422
Timestamp: 2026-07-06T15:26:18.398Z
Learning: In this repo’s Python code under src/**, don’t treat differences in string concatenation style as a style inconsistency when Black has effectively forced (or made clearer) use of explicit `+` string concatenation in multi-line logger/string expressions. If adjacent-literal implicit concatenation is avoided/changed specifically to accommodate Black’s formatting in these call sites, accept the `+` usage and don’t recommend converting it solely for consistency with nearby blocks that use implicit concatenation.

Applied to files:

• src/models/common/query.py
• src/llama_stack_configuration.py

📚 Learning: 2026-07-17T19:25:05.325Z

Learnt from: Jdubrick
Repo: lightspeed-core/lightspeed-stack PR: 2166
File: src/utils/saved_prompts.py:129-157
Timestamp: 2026-07-17T19:25:05.325Z
Learning: For any endpoint that handles saved prompts and calls `src/utils/saved_prompts.py::create_saved_prompt`, treat the endpoint as the validation boundary. Before calling `create_saved_prompt`, validate the incoming saved-prompt name and content, specifically using `validate_saved_prompt_name` and then persist (store) the normalized value it returns. Do not call `create_saved_prompt` with unvalidated/raw name/content.

Applied to files:

• src/models/common/query.py
• src/llama_stack_configuration.py

🔇 Additional comments (4)

src/llama_stack_configuration.py (1)

770-777: LGTM!

src/models/common/query.py (3)

138-142: Align the legacy-client fallback documentation.

Line 142 still says that a legacy filter-only object uses the server default. src/utils/vector_search.py resolves mode=None from configuration.okp.search_mode, then SOLR_VECTOR_SEARCH_DEFAULT_MODE. Update Line 142 to describe that precedence. This remains the documentation mismatch reported in the previous review.

---

147-154: LGTM!

---

210-210: LGTM!

[maysunfaisal]

/retest