feat: add Requesty as LLM provider by Thibault00 · Pull Request #1245 · Tracer-Cloud/opensre

Thibault00 · 2026-05-03T15:20:42Z

Describe the changes you have made in this PR -

Hey! We're Requesty, an OpenAI-compatible LLM gateway (300+ models, fallback routing, caching, cost optimization). We've been using OpenSRE for our SRE workflows and wanted to add native support for routing LLM calls through Requesty.

The integration follows the exact same pattern as OpenRouter since we're also OpenAI-compatible. Every request sent through Requesty includes an X-Title: OpenSRE header so traffic is identifiable on both sides. Would love to explore a deeper connection between the two projects if there's interest!

What changed:

app/config.py — added Requesty model constants (anthropic/claude-sonnet-4-6 for both reasoning and toolcall), base URL, LLMModelConfig, LLMProvider literal, LLMSettings fields, validator, and from_env() env var resolution
app/services/llm_client.py — added default_headers param to OpenAILLMClient (forwarded to the OpenAI constructor), and a new elif provider == "requesty" branch that passes X-Title: OpenSRE
app/cli/wizard/config.py — added REQUESTY_MODELS (5 curated models: Claude Opus 4.7, Claude Sonnet 4.6, GPT-5.5, Gemini 3.1 Pro, Gemini 3.1 Flash-Lite) and a ProviderOption entry in SUPPORTED_PROVIDERS
app/cli/wizard/validation.py — added base URL resolution for credential validation
app/cli/commands/doctor.py — opensre health now recognizes REQUESTY_API_KEY
app/cli/interactive_shell/agent_actions.py — runtime provider switching support
.env.example — documented REQUESTY_API_KEY, REQUESTY_REASONING_MODEL, REQUESTY_TOOLCALL_MODEL

Demo/Screenshot for feature changes and bug fixes -

$ REQUESTY_API_KEY=test-key LLM_PROVIDER=requesty python3 -c "
from app.config import REQUESTY_BASE_URL, REQUESTY_LLM_CONFIG, LLMSettings
s = LLMSettings.from_env()
print(f'Provider: {s.provider}')
print(f'Reasoning: {s.requesty_reasoning_model}')
print(f'Toolcall: {s.requesty_toolcall_model}')
print(f'Base URL: {REQUESTY_BASE_URL}')
"
Provider: requesty
Reasoning: anthropic/claude-sonnet-4-6
Toolcall: anthropic/claude-sonnet-4-6
Base URL: https://router.requesty.ai/v1

Local checks on changed files all pass:

ruff check (6 changed files)  -> All checks passed!
ruff format --check            -> 6 files already formatted
mypy                           -> 0 errors in changed files
pytest                         -> 3815 passed

Code Understanding and AI Usage

Did you use AI assistance (ChatGPT, Claude, Copilot, etc.) to write any part of this code?

No, I wrote all the code myself
Yes, I used AI assistance (continue below)

If you used AI assistance:

I have reviewed every single line of the AI-generated code
I can explain the purpose and logic of each function/component I added
I have tested edge cases and understand how the code handles them
I have modified the AI output to follow this project's coding standards and conventions

Explain your implementation approach:

Requesty is OpenAI-compatible so it slots right into the existing OpenAILLMClient path, same as OpenRouter, Gemini, and NVIDIA. The only new concept is the default_headers parameter on OpenAILLMClient which lets us pass X-Title: OpenSRE without touching other providers. The OpenAI Python SDK already supports default_headers natively so this is just threading it through.

Followed the exact same pattern as the OpenRouter integration across all 7 files: constants, config model, client factory, wizard, validation, doctor check, and interactive shell. No new abstractions, no new dependencies.

Checklist before requesting a review

I have added proper PR title and linked to the issue
I have performed a self-review of my code
I can explain the purpose of every function, class, and logic block I added
I understand why my changes work and have tested them thoroughly
I have considered potential edge cases and how my code handles them
If it is a core feature, I have added thorough tests
My code follows the project's style guidelines and conventions

Note: Please check Allow edits from maintainers if you would like us to assist in the PR.

Add Requesty as an LLM provider option in config: - Model constants (anthropic/claude-sonnet-4-6 reasoning, anthropic/claude-haiku-4-5 toolcall) - Base URL (https://router.requesty.ai/v1) - LLMModelConfig, LLMProvider Literal, LLMSettings fields

Both default to anthropic/claude-sonnet-4-6.

Accept optional default_headers dict and forward it to the OpenAI constructor. This enables per-provider custom headers like X-Title.

Route LLM_PROVIDER=requesty through OpenAILLMClient with router.requesty.ai/v1 base URL and X-Title: OpenSRE header.

Add REQUESTY_MODELS model catalog, ProviderOption entry in SUPPORTED_PROVIDERS, and validation base URL resolution.

Register REQUESTY_API_KEY in health check and add requesty to the interactive shell provider name set.

Add requesty_api_key to provider_to_key and env_var maps in the model validator, and add env var resolution in from_env().

greptile-apps · 2026-05-03T15:22:28Z

Greptile Summary

This PR adds Requesty as a new OpenAI-compatible LLM gateway provider, following the same integration pattern used by OpenRouter, Gemini, and NVIDIA. Changes span config constants, LLMSettings fields, from_env resolution, the setup wizard, validation, doctor command, and OpenAILLMClient (which gains a default_headers parameter used to send X-Title: OpenSRE to Requesty).

Confidence Score: 4/5

Safe to merge with minor cleanup; no runtime-breaking issues found.

Only P2 findings: a duplicate label in the wizard model list and a misplaced dict entry in from_env. Core integration logic is consistent with existing provider patterns.

app/cli/wizard/config.py — duplicate "Claude Sonnet 4.6 (via Requesty)" label in REQUESTY_MODELS.

Important Files Changed

Filename	Overview
app/cli/wizard/config.py	Adds REQUESTY_MODELS and ProviderOption; two entries in REQUESTY_MODELS share the same "Claude Sonnet 4.6 (via Requesty)" label with different underlying model paths.
app/config.py	Adds Requesty constants, LLMSettings fields, and from_env resolution — structurally mirrors other providers, but requesty_api_key is placed out of order with other API-key entries.
app/services/llm_client.py	Adds default_headers support to OpenAILLMClient and wires Requesty provider with X-Title header; consistent with existing provider patterns.
app/cli/wizard/validation.py	Adds Requesty base URL lookup to _get_provider_base_url, consistent with other provider branches.
app/cli/commands/doctor.py	Adds requesty → REQUESTY_API_KEY to the provider health-check map.
app/cli/interactive_shell/agent_actions.py	Adds "requesty" to the frozen set of known LLM provider names.
.env.example	Documents REQUESTY_API_KEY, REQUESTY_REASONING_MODEL, and REQUESTY_TOOLCALL_MODEL env vars.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[LLM_PROVIDER=requesty] --> B[LLMSettings.from_env]
    B --> C{provider == requesty?}
    C -->|Yes| D[_create_llm_client]
    D --> E[Load REQUESTY_LLM_CONFIG & REQUESTY_BASE_URL]
    E --> F[OpenAILLMClient\nbase_url=router.requesty.ai/v1\ndefault_headers: X-Title OpenSRE]
    F --> G[OpenAI SDK to Requesty Gateway]
    G --> H{Route by model prefix}
    H --> I[anthropic/ to Claude]
    H --> J[bedrock/ to AWS Bedrock]
    H --> K[openai/ to GPT]
    H --> L[vertex/ to Gemini]

_{Reviews (1): Last reviewed commit: "fix: add Requesty to LLMSettings validat..." | Re-trigger Greptile}

- Disambiguate duplicate "Claude Sonnet 4.6" labels in REQUESTY_MODELS by adding "Bedrock" qualifier to bedrock/ model entries - Move requesty_api_key to the API key group in from_env() for consistency with other providers

gitsofaryan · 2026-05-03T17:28:53Z

@Thibault00 CI is failing and please address greptile issues!

VaibhavUpreti

Awesome @Thaibat - really appreciate the PR! Great to hear you’ve been using OpenSRE in your SRE workflows.

We’re adding a lot more product improvements and features soon, so excited to have Requesty support in the project.

I'll take this branch from here.

github-actions · 2026-05-03T17:35:02Z

🚀 Houston, we have a merge. @Thibault00 your PR is in orbit. Thanks for launching this one!

👋 Join us on Discord - OpenSRE : hang out, contribute, or hunt for features and issues. Everyone's welcome.

Thibault00 added 9 commits May 3, 2026 16:08

fix: use same model for both Requesty reasoning and toolcall

6a5b3e0

Both default to anthropic/claude-sonnet-4-6.

feat: add default_headers support to OpenAILLMClient

01c7105

Accept optional default_headers dict and forward it to the OpenAI constructor. This enables per-provider custom headers like X-Title.

feat: wire Requesty provider in LLM client factory

7977e50

Route LLM_PROVIDER=requesty through OpenAILLMClient with router.requesty.ai/v1 base URL and X-Title: OpenSRE header.

feat: add Requesty to CLI wizard and onboarding

7bb82e4

Add REQUESTY_MODELS model catalog, ProviderOption entry in SUPPORTED_PROVIDERS, and validation base URL resolution.

feat: add Requesty to doctor check and interactive shell

8e7e5c2

Register REQUESTY_API_KEY in health check and add requesty to the interactive shell provider name set.

docs: add Requesty env vars to .env.example

d563063

style: format llm_client.py long lines

f4a36c8

fix: add Requesty to LLMSettings validator and from_env

194f21c

Add requesty_api_key to provider_to_key and env_var maps in the model validator, and add env var resolution in from_env().

Thibault00 changed the title ~~Feat/add requesty llm provider~~ feat: add Requesty as LLM provider May 3, 2026

greptile-apps Bot reviewed May 3, 2026

View reviewed changes

Comment thread app/cli/wizard/config.py Outdated

Comment thread app/config.py Outdated

fix: address review feedback on Requesty integration

9a96c02

- Disambiguate duplicate "Claude Sonnet 4.6" labels in REQUESTY_MODELS by adding "Bedrock" qualifier to bedrock/ model entries - Move requesty_api_key to the API key group in from_env() for consistency with other providers

VaibhavUpreti approved these changes May 3, 2026

View reviewed changes

VaibhavUpreti merged commit ad057f9 into Tracer-Cloud:main May 3, 2026
9 of 10 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: add Requesty as LLM provider#1245

feat: add Requesty as LLM provider#1245
VaibhavUpreti merged 10 commits intoTracer-Cloud:mainfrom
Thibault00:feat/add-requesty-llm-provider

Thibault00 commented May 3, 2026 •

edited

Loading

Uh oh!

greptile-apps Bot commented May 3, 2026

Uh oh!

Uh oh!

Uh oh!

gitsofaryan commented May 3, 2026

Uh oh!

VaibhavUpreti left a comment

Uh oh!

Uh oh!

github-actions Bot commented May 3, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

Thibault00 commented May 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Describe the changes you have made in this PR -

Demo/Screenshot for feature changes and bug fixes -

Code Understanding and AI Usage

Checklist before requesting a review

Uh oh!

greptile-apps Bot commented May 3, 2026

Greptile Summary

Confidence Score: 4/5

Important Files Changed

Flowchart

Uh oh!

Uh oh!

Uh oh!

gitsofaryan commented May 3, 2026

Uh oh!

VaibhavUpreti left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

github-actions Bot commented May 3, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Thibault00 commented May 3, 2026 •

edited

Loading