Add sambanova provider

[luisfucros]

Add SambaNova provider. SambaNova provides high-performance infrastructure and cloud services for running large language models (LLMs) and Inference APIs for models families like DeepSeek, Llama, GPT and MiniMax.

[qodo-free-for-open-source-projects]

Review Summary by Qodo

Add SambaNova provider integration and documentation

✨ Enhancement

Walkthroughs

Description

• Add SambaNova provider support with four model configurations
• Integrate SambaNova API key handling in litellm AI handler
• Add SambaNova documentation and configuration examples
• Update context window limits for SambaNova models

Diagram

flowchart LR
  A["SambaNova Provider"] --> B["Model Context Windows"]
  A --> C["API Key Configuration"]
  A --> D["Documentation & Setup"]
  B --> E["4 Models Added"]
  C --> F["litellm Handler"]
  D --> G["User Guide"]

Loading

File Changes

1. pr_agent/algo/__init__.py ⚙️ Configuration changes +4/-0

Add SambaNova models to context window configuration

• Added four SambaNova model entries to context window mapping
• Models include MiniMax-M2.5, Meta-Llama-3.3-70B-Instruct, gpt-oss-120b, and DeepSeek-V3.1
• All models configured with 128000 token context window limit

pr_agent/algo/init.py

---

2. pr_agent/algo/ai_handlers/litellm_ai_handler.py ✨ Enhancement +2/-0

Integrate SambaNova API key handling

• Added SambaNova API key initialization in the handler constructor
• Checks for SAMBANOVA.KEY setting and configures litellm.api_key
• Follows same pattern as existing providers like Groq and xAI

pr_agent/algo/ai_handlers/litellm_ai_handler.py

---

3. docs/docs/usage-guide/changing_a_model.md 📝 Documentation +21/-6

Add SambaNova setup guide and fix formatting

• Added new SambaNova section with configuration examples
• Provided instructions for setting model and API key in configuration files
• Included link to SambaNova API key acquisition page
• Fixed indentation and formatting issues in existing documentation
• Corrected trailing whitespace in multiple sections

docs/docs/usage-guide/changing_a_model.md

---

View more (1)

4. pr_agent/settings/.secrets_template.toml ⚙️ Configuration changes +3/-0

Add SambaNova configuration to secrets template

• Added new [sambanova] section to secrets template
• Included placeholder for SambaNova API key with acquisition link
• Positioned between groq and xai sections for logical organization

pr_agent/settings/.secrets_template.toml

---

[qodo-free-for-open-source-projects]

Code Review by Qodo

🐞 Bugs (1)   📘 Rule violations (1)   📎 Requirement gaps (0)

🐞\ ☼ Reliability (1)

📘\ ⚙ Maintainability (1) ⭐ New (1)

1. MkDocs admonition not indented ☑ 📘 ☼

Description

The !!! note "Model-specific environment variables" block content is not indented, which can break
MkDocs/MkDocs-Material admonition rendering and potentially fail the docs build.

Code

docs/docs/usage-guide/changing_a_model.md[R16-18]

!!! note "Model-specific environment variables"
-    See [litellm documentation](https://litellm.vercel.app/docs/proxy/quick_start#supported-llms) for the environment variables needed per model, as they may vary and change over time. Our documentation per-model may not always be up-to-date with the latest changes.
-    Failing to set the needed keys of a specific model will usually result in litellm not identifying the model type, and failing to utilize it.
+See [litellm documentation](https://litellm.vercel.app/docs/proxy/quick_start#supported-llms) for the environment variables needed per model, as they may vary and change over time. Our documentation per-model may not always be up-to-date with the latest changes.
+Failing to set the needed keys of a specific model will usually result in litellm not identifying the model type, and failing to utilize it.

Evidence

PR Compliance ID 11 requires docs Markdown under docs/ to follow MkDocs conventions. The added
lines immediately after the admonition marker are not indented as required for admonition content.

AGENTS.md
docs/docs/usage-guide/changing_a_model.md[16-18]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The Markdown admonition content is not indented under the `!!! note` marker, which can break MkDocs admonition parsing/rendering.
## Issue Context
MkDocs-style admonitions require the content block to be indented relative to the `!!!` line.
## Fix Focus Areas
- docs/docs/usage-guide/changing_a_model.md[16-18]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

2. Second admonition line unindented ☑ 📘 ☼

Description

The !!! note "Local models vs commercial models" block has its first content line unindented while
the rest is indented, which can break MkDocs admonition formatting.

Code

docs/docs/usage-guide/changing_a_model.md[R109-112]

!!! note "Local models vs commercial models"
-    PR-Agent is compatible with almost any AI model, but analyzing complex code repositories and pull requests requires a model specifically optimized for code analysis.
+PR-Agent is compatible with almost any AI model, but analyzing complex code repositories and pull requests requires a model specifically optimized for code analysis.
Commercial models such as GPT-5, Claude Sonnet, and Gemini have demonstrated robust capabilities in generating structured output for code analysis tasks with large input. In contrast, most open-source models currently available (as of January 2025) face challenges with these complex tasks.

Evidence

PR Compliance ID 11 requires MkDocs-compliant Markdown in docs/. The first line of admonition
content is not indented under the !!! note marker, making the block syntactically inconsistent.

AGENTS.md
docs/docs/usage-guide/changing_a_model.md[109-112]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The admonition content immediately following `!!! note "Local models vs commercial models"` is not consistently indented, which can break MkDocs admonition parsing.
## Issue Context
In MkDocs-style admonitions, all content lines belonging to the admonition must be indented under the `!!!` directive.
## Fix Focus Areas
- docs/docs/usage-guide/changing_a_model.md[109-112]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

3. Wrong provider key injected 🐞 ☼

Description

LiteLLMAIHandler.__init__ now overwrites litellm.api_key with the SambaNova key, and
chat_completion injects this single global key into every LiteLLM call, so mixed-provider
fallback_models can be invoked with the wrong API key and fail authentication.

Code

pr_agent/algo/ai_handlers/litellm_ai_handler.py[R74-75]

+        if get_settings().get("SAMBANOVA.KEY", None):
+            litellm.api_key = get_settings().sambanova.key

Evidence

During initialization, multiple providers write to the same global litellm.api_key sequentially
(Groq first, then SambaNova, then others). Later, chat_completion unconditionally forwards
litellm.api_key as api_key for every acompletion call. Since PR-Agent can iterate across
config.model + config.fallback_models, a fallback from a different provider can be attempted
while still sending the last-initialized provider’s key, causing auth failures. The existing unit
test explicitly documents this overwrite behavior when multiple providers are configured.

pr_agent/algo/ai_handlers/litellm_ai_handler.py[68-90]
pr_agent/algo/ai_handlers/litellm_ai_handler.py[412-416]
pr_agent/algo/pr_processing.py[320-354]
tests/unittest/test_litellm_api_key_guard.py[276-333]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`LiteLLMAIHandler` uses `litellm.api_key` as a single global, but PR-Agent can call multiple providers via `fallback_models`. With the new SambaNova assignment, whichever provider runs last in `__init__` wins, and `chat_completion` forwards that key for *all* models.
### Issue Context
- `__init__` sets `litellm.api_key` for multiple providers (Groq, SambaNova, xAI, Ollama, OpenRouter, Azure AD, etc.).
- `chat_completion` injects `kwargs["api_key"] = litellm.api_key` for every call.
- `retry_with_fallback_models()` may invoke different providers in a single run.
### Fix Focus Areas
- pr_agent/algo/ai_handlers/litellm_ai_handler.py[68-90]
- pr_agent/algo/ai_handlers/litellm_ai_handler.py[412-416]
- pr_agent/algo/pr_processing.py[320-354]
### Suggested approach
1. Stop relying on a single global `litellm.api_key` for multiple providers.
2. In `chat_completion`, derive provider from `model` (e.g., prefix before `/`) and set `kwargs["api_key"]` from the corresponding settings (e.g., `get_settings().groq.key`, `get_settings().sambanova.key`, `get_settings().xai.key`, etc.).
3. Only inject `api_key` when you have a provider-specific match; otherwise let LiteLLM handle OpenAI/Azure/Anthropic keys via their dedicated fields/env vars.
4. Add/extend a unit test covering Groq + SambaNova configured simultaneously, then calling both a `groq/...` and `sambanova/...` model and verifying the forwarded `api_key` differs appropriately.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

4. Mixed access for SAMBANOVA.KEY 📘 ⚙ ⭐

Description

The new SambaNova key handling checks get_settings().get("SAMBANOVA.KEY") but then reads
get_settings().sambanova.key, mixing configuration access patterns/casing. This violates the
requirement for consistent configuration access and increases the risk of runtime errors or config
drift across providers.

Code

pr_agent/algo/ai_handlers/litellm_ai_handler.py[R74-75]

+        if get_settings().get("SAMBANOVA.KEY", None):
+            litellm.api_key = get_settings().sambanova.key

Evidence

PR Compliance ID 16 requires a consistent configuration access pattern and stable key paths; the
added SambaNova logic mixes .get("SAMBANOVA.KEY") with attribute access .sambanova.key on the
same setting.

pr_agent/algo/ai_handlers/litellm_ai_handler.py[74-75]
Best Practice: Learned patterns

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The SambaNova API key retrieval mixes Dynaconf access patterns (`get_settings().get("SAMBANOVA.KEY")` vs `get_settings().sambanova.key`) for the same configuration value.

## Issue Context
PR compliance requires using a single consistent access pattern and stable key paths to avoid subtle configuration bugs.

## Fix Focus Areas
- pr_agent/algo/ai_handlers/litellm_ai_handler.py[74-75]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

ⓘ The new review experience is currently in Beta. Learn more

Previous review results

Review updated until commit ba4048f

Results up to commit N/A

🐞 Bugs (1)  
📘 Rule violations (0)  
📎 Requirement gaps (0)

1. MkDocs admonition not indented ☑ 📘 ☼

Description

The !!! note "Model-specific environment variables" block content is not indented, which can break
MkDocs/MkDocs-Material admonition rendering and potentially fail the docs build.

Code

docs/docs/usage-guide/changing_a_model.md[R16-18]

!!! note "Model-specific environment variables"
-    See [litellm documentation](https://litellm.vercel.app/docs/proxy/quick_start#supported-llms) for the environment variables needed per model, as they may vary and change over time. Our documentation per-model may not always be up-to-date with the latest changes.
-    Failing to set the needed keys of a specific model will usually result in litellm not identifying the model type, and failing to utilize it.
+See [litellm documentation](https://litellm.vercel.app/docs/proxy/quick_start#supported-llms) for the environment variables needed per model, as they may vary and change over time. Our documentation per-model may not always be up-to-date with the latest changes.
+Failing to set the needed keys of a specific model will usually result in litellm not identifying the model type, and failing to utilize it.

Evidence

PR Compliance ID 11 requires docs Markdown under docs/ to follow MkDocs conventions. The added
lines immediately after the admonition marker are not indented as required for admonition content.

AGENTS.md
docs/docs/usage-guide/changing_a_model.md[16-18]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The Markdown admonition content is not indented under the `!!! note` marker, which can break MkDocs admonition parsing/rendering.
## Issue Context
MkDocs-style admonitions require the content block to be indented relative to the `!!!` line.
## Fix Focus Areas
- docs/docs/usage-guide/changing_a_model.md[16-18]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

2. Second admonition line unindented ☑ 📘 ☼

Description

The !!! note "Local models vs commercial models" block has its first content line unindented while
the rest is indented, which can break MkDocs admonition formatting.

Code

docs/docs/usage-guide/changing_a_model.md[R109-112]

!!! note "Local models vs commercial models"
-    PR-Agent is compatible with almost any AI model, but analyzing complex code repositories and pull requests requires a model specifically optimized for code analysis.
+PR-Agent is compatible with almost any AI model, but analyzing complex code repositories and pull requests requires a model specifically optimized for code analysis.
 Commercial models such as GPT-5, Claude Sonnet, and Gemini have demonstrated robust capabilities in generating structured output for code analysis tasks with large input. In contrast, most open-source models currently available (as of January 2025) face challenges with these complex tasks.

Evidence

PR Compliance ID 11 requires MkDocs-compliant Markdown in docs/. The first line of admonition
content is not indented under the !!! note marker, making the block syntactically inconsistent.

AGENTS.md
docs/docs/usage-guide/changing_a_model.md[109-112]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The admonition content immediately following `!!! note "Local models vs commercial models"` is not consistently indented, which can break MkDocs admonition parsing.
## Issue Context
In MkDocs-style admonitions, all content lines belonging to the admonition must be indented under the `!!!` directive.
## Fix Focus Areas
- docs/docs/usage-guide/changing_a_model.md[109-112]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

3. Wrong provider key injected 🐞 ☼

Description

LiteLLMAIHandler.__init__ now overwrites litellm.api_key with the SambaNova key, and
chat_completion injects this single global key into every LiteLLM call, so mixed-provider
fallback_models can be invoked with the wrong API key and fail authentication.

Code

pr_agent/algo/ai_handlers/litellm_ai_handler.py[R74-75]

+        if get_settings().get("SAMBANOVA.KEY", None):
+            litellm.api_key = get_settings().sambanova.key

Evidence

During initialization, multiple providers write to the same global litellm.api_key sequentially
(Groq first, then SambaNova, then others). Later, chat_completion unconditionally forwards
litellm.api_key as api_key for every acompletion call. Since PR-Agent can iterate across
config.model + config.fallback_models, a fallback from a different provider can be attempted
while still sending the last-initialized provider’s key, causing auth failures. The existing unit
test explicitly documents this overwrite behavior when multiple providers are configured.

pr_agent/algo/ai_handlers/litellm_ai_handler.py[68-90]
pr_agent/algo/ai_handlers/litellm_ai_handler.py[412-416]
pr_agent/algo/pr_processing.py[320-354]
tests/unittest/test_litellm_api_key_guard.py[276-333]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`LiteLLMAIHandler` uses `litellm.api_key` as a single global, but PR-Agent can call multiple providers via `fallback_models`. With the new SambaNova assignment, whichever provider runs last in `__init__` wins, and `chat_completion` forwards that key for *all* models.
### Issue Context
- `__init__` sets `litellm.api_key` for multiple providers (Groq, SambaNova, xAI, Ollama, OpenRouter, Azure AD, etc.).
- `chat_completion` injects `kwargs["api_key"] = litellm.api_key` for every call.
- `retry_with_fallback_models()` may invoke different providers in a single run.
### Fix Focus Areas
- pr_agent/algo/ai_handlers/litellm_ai_handler.py[68-90]
- pr_agent/algo/ai_handlers/litellm_ai_handler.py[412-416]
- pr_agent/algo/pr_processing.py[320-354]
### Suggested approach
1. Stop relying on a single global `litellm.api_key` for multiple providers.
2. In `chat_completion`, derive provider from `model` (e.g., prefix before `/`) and set `kwargs["api_key"]` from the corresponding settings (e.g., `get_settings().groq.key`, `get_settings().sambanova.key`, `get_settings().xai.key`, etc.).
3. Only inject `api_key` when you have a provider-specific match; otherwise let LiteLLM handle OpenAI/Azure/Anthropic keys via their dedicated fields/env vars.
4. Add/extend a unit test covering Groq + SambaNova configured simultaneously, then calling both a `groq/...` and `sambanova/...` model and verifying the forwarded `api_key` differs appropriately.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

[qodo-free-for-open-source-projects]

3. Wrong provider key injected 🐞 Bug ☼ Reliability

LiteLLMAIHandler.__init__ now overwrites litellm.api_key with the SambaNova key, and
chat_completion injects this single global key into every LiteLLM call, so mixed-provider
fallback_models can be invoked with the wrong API key and fail authentication.

Agent Prompt

### Issue description
`LiteLLMAIHandler` uses `litellm.api_key` as a single global, but PR-Agent can call multiple providers via `fallback_models`. With the new SambaNova assignment, whichever provider runs last in `__init__` wins, and `chat_completion` forwards that key for *all* models.

### Issue Context
- `__init__` sets `litellm.api_key` for multiple providers (Groq, SambaNova, xAI, Ollama, OpenRouter, Azure AD, etc.).
- `chat_completion` injects `kwargs["api_key"] = litellm.api_key` for every call.
- `retry_with_fallback_models()` may invoke different providers in a single run.

### Fix Focus Areas
- pr_agent/algo/ai_handlers/litellm_ai_handler.py[68-90]
- pr_agent/algo/ai_handlers/litellm_ai_handler.py[412-416]
- pr_agent/algo/pr_processing.py[320-354]

### Suggested approach
1. Stop relying on a single global `litellm.api_key` for multiple providers.
2. In `chat_completion`, derive provider from `model` (e.g., prefix before `/`) and set `kwargs["api_key"]` from the corresponding settings (e.g., `get_settings().groq.key`, `get_settings().sambanova.key`, `get_settings().xai.key`, etc.).
3. Only inject `api_key` when you have a provider-specific match; otherwise let LiteLLM handle OpenAI/Azure/Anthropic keys via their dedicated fields/env vars.
4. Add/extend a unit test covering Groq + SambaNova configured simultaneously, then calling both a `groq/...` and `sambanova/...` model and verifying the forwarded `api_key` differs appropriately.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

[qodo-free-for-open-source-projects]

Persistent review updated to latest commit 21901fc

[qodo-free-for-open-source-projects]

Persistent review updated to latest commit ba4048f