Add configurable LLM API URL and model allowlists

[davidmz]

Enables configuration of custom LLM model allowlists and OpenAI-compatible API endpoints through environment variables. Existing provider and model behavior remains unchanged when the new variables are not set.

Description

• Adds LLM_MODELS as an optional comma-separated model allowlist.
• Adds LLM_API_BASE for custom OpenAI-compatible API endpoints.
• Preserves each provider's built-in model allowlist when LLM_MODELS is empty.
• Passes the configured API base directly to the OpenAI client.
• Adds the variables to Compose configuration and environment templates.
• Documents configuration for CLI and AIO deployments, including God Mode limitations.
• Adds unit coverage for default behavior, custom allowlists, validation, and API base handling.

Example:

LLM_PROVIDER=openai
LLM_MODEL=gpt-5-mini
LLM_MODELS=gpt-5-mini,gpt-5.4-nano,qwen3-235b
LLM_API_BASE=https://example.com/v1

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• Feature (non-breaking change which adds functionality)
• Improvement (change that would cause existing functionality to not work as expected)
• Code refactoring
• Performance improvements
• Documentation update

Screenshots and Media (if applicable)

Not applicable. The new settings are configured through environment variables and are not exposed in God Mode.

Test Scenarios

• Verified that the built-in provider allowlist remains active when LLM_MODELS is unset.
• Verified parsing of a comma-separated custom model allowlist.
• Verified that models outside the configured allowlist are rejected.
• Verified default OpenAI client creation when LLM_API_BASE is unset.
• Verified that a custom LLM_API_BASE is passed to the OpenAI client.
• Ran the focused unit test file: 5 tests passed.
• Ran Ruff checks for the changed Python files.
• Validated the Docker Compose configuration with custom LLM values.
• Ran git diff --check.

The full test suite was not run.

References

None.

Summary by CodeRabbit

• New Features

  • Added configurable LLM provider, model, model allowlist, API key, and API base URL settings.
  • Added support for custom API endpoints and model validation.
  • Improved handling of blank or invalid configuration values.

• Documentation

  • Added self-hosted LLM setup guidance, configuration examples, restart requirements, and platform-specific limitations.

[Copilot]

Pull request overview

Adds environment-driven support for (1) custom LLM model allowlists and (2) custom OpenAI-compatible API base URLs, while keeping existing provider defaults when overrides are not configured. This primarily updates the API’s external LLM integration helpers and propagates new env vars into deployment templates/docs.

Changes:

• Extend get_llm_config() to read LLM_MODELS and LLM_API_BASE, validate models against either provider defaults or the override list, and return api_base.
• Pass api_base through to the OpenAI client (base_url=...) in get_llm_response().
• Add unit tests and update CLI/AIO deployment env templates + docs to include the new configuration knobs.

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 9 comments.

Show a summary per file

File	Description
deployments/cli/community/variables.env	Adds LLM env knobs for CLI deployments (provider/model/models/api base).
deployments/cli/community/README.md	Documents new LLM env configuration and God Mode limitations for CLI deploy.
deployments/cli/community/docker-compose.yml	Wires LLM env vars into the shared app environment passed to services.
deployments/aio/community/variables.env	Adds LLM env knobs for AIO deployments.
deployments/aio/community/README.md	Documents new LLM env configuration for AIO deploy.
apps/api/plane/tests/unit/views/test_external_llm.py	Adds unit coverage for default vs overridden LLM config + API base handling.
apps/api/plane/app/views/external/base.py	Implements LLM_MODELS allowlist override and passes LLM_API_BASE to OpenAI client.
apps/api/.env.example	Adds LLM env knobs to backend env template.
.env.example	Adds LLM env knobs to root env template (alongside deprecated OpenAI/GPT vars).

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 660175e3-1464-474a-bc5b-9a8d17643ff1

📥 Commits

Reviewing files that changed from the base of the PR and between a46321f and 289288c.

📒 Files selected for processing (1)

• apps/api/plane/app/views/external/base.py

🚧 Files skipped from review as they are similar to previous changes (1)

• apps/api/plane/app/views/external/base.py

---

📝 Walkthrough

Walkthrough

LLM configuration now supports model lists, optional API base URLs, model validation, and four-value error results. Integration endpoints pass the resolved API base URL to response generation. Deployment templates, documentation, and tests cover the new settings.

Changes

LLM Configuration

Layer / File(s)	Summary
Runtime configuration and response propagation apps/api/plane/app/views/external/base.py, apps/api/plane/tests/unit/views/test_external_llm.py	LLM configuration resolves model overrides and optional API base URLs. It validates models and returns four-value error results. Both integration endpoints pass the API base URL to response generation. Unit tests cover configuration and client initialization.
Deployment configuration and documentation .env.example, apps/api/.env.example, deployments/aio/community/*, deployments/cli/community/*	Environment templates and Docker Compose define the LLM variables. Deployment documentation describes defaults, overrides, initialization, restart requirements, and updated sample logs.

Estimated code review effort: 2 (Simple) | ~10 minutes

Suggested reviewers: dheeru0198

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly summarizes the main changes: configurable LLM API URLs and model allowlists.
Description check	✅ Passed	The description includes all required sections and provides clear change details, test scenarios, and deployment documentation scope.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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.}