feat: add EvoLink LLM provider

[EvoLinkAI]

Related Issue

Closes #194

Description

Adds EvoLink as a first-class LLM provider for Director's existing BaseLLM abstraction.

EvoLink's language API is OpenAI-compatible, so this PR keeps the implementation small by reusing the existing OpenAI chat-completions formatting and response parsing while adding EvoLink-specific configuration defaults:

• EVOLINK_API_KEY
• EVOLINK_API_BASE=https://direct.evolink.ai/v1
• EVOLINK_CHAT_MODEL=gpt-5.2
• DEFAULT_LLM=evolink

This also adds MkDocs documentation, updates the sample env file, and covers provider configuration/default selection with unit tests.

Testing

env -u all_proxy -u http_proxy -u https_proxy -u ALL_PROXY -u HTTP_PROXY -u HTTPS_PROXY \
  .venv/bin/ruff check director/constants.py director/llm/__init__.py director/llm/evolink.py tests/llm/test_evolink.py
# All checks passed!

env -u all_proxy -u http_proxy -u https_proxy -u ALL_PROXY -u HTTP_PROXY -u HTTPS_PROXY \
  .venv/bin/black --check director/llm/evolink.py director/llm/__init__.py director/constants.py tests/llm/test_evolink.py
# All done! 4 files would be left unchanged.

env -u all_proxy -u http_proxy -u https_proxy -u ALL_PROXY -u HTTP_PROXY -u HTTPS_PROXY \
  .venv/bin/python -m pytest tests/llm/test_evolink.py -q
# 4 passed, 10 warnings in 0.61s

env -u all_proxy -u http_proxy -u https_proxy -u ALL_PROXY -u HTTP_PROXY -u HTTPS_PROXY \
  .venv/bin/python -m py_compile director/constants.py director/llm/__init__.py director/llm/evolink.py tests/llm/test_evolink.py
# passed

Real EvoLink compatibility check:

curl -X POST https://direct.evolink.ai/v1/chat/completions \
  -H "Authorization: Bearer $EVOLINK_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"model":"gpt-5.2","messages":[{"role":"user","content":"Reply with exactly: Director EvoLink OK"}],"max_tokens":32}'
# HTTP 200
# Director EvoLink OK

Director provider path:

from dotenv import load_dotenv
load_dotenv('/path/to/.env', override=True)
from director.llm.evolink import Evolink, EvolinkConfig
model = Evolink(EvolinkConfig(max_tokens=32, temperature=0))
response = model.chat_completions([
    {'role': 'user', 'content': 'Reply with exactly: Director EvoLink provider OK'}
])
print(response.status, response.content, model.chat_model, model.api_base)
# status= 1
# content= Director EvoLink provider OK
# model= gpt-5.2
# base= https://direct.evolink.ai/v1

Attempted docs validation:

backend/.venv/bin/mkdocs build --strict
# blocked before page build by dev dependency resolution:
# ModuleNotFoundError: No module named 'mkdocstrings.extension'

Summary by CodeRabbit

Release Notes

• New Features

  • Added EvoLink as a supported LLM integration option alongside OpenAI.
  • EvoLink can be configured via environment variables (EVOLINK_API_KEY, EVOLINK_API_BASE, EVOLINK_CHAT_MODEL).

• Documentation

  • Added EvoLink integration guide with setup instructions and configuration details.
  • Updated LLM documentation to reflect available integration options.

• Tests

  • Added test coverage for EvoLink configuration and default LLM selection.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 515aa11b-4f95-4b9b-80f1-66bdb323eef4

📥 Commits

Reviewing files that changed from the base of the PR and between 70e0b3d and 614f2bb.

📒 Files selected for processing (8)

• backend/.env.sample
• backend/README.md
• backend/director/constants.py
• backend/director/llm/__init__.py
• backend/director/llm/evolink.py
• backend/tests/llm/test_evolink.py
• docs/llm/evolink.md
• mkdocs.yml

---

📝 Walkthrough

Walkthrough

Adds EvoLink as an OpenAI-compatible LLM provider. Introduces LLMType.EVOLINK and EnvPrefix.EVOLINK_ constants, a new evolink.py module with EvolinkConfig and Evolink classes, get_default_llm() selection logic, unit tests, updated env sample, README, and a new documentation page.

Changes

EvoLink LLM Provider

Layer / File(s)	Summary
Enum constants and EvoLink module backend/director/constants.py, backend/director/llm/evolink.py	LLMType.EVOLINK and EnvPrefix.EVOLINK_ added to constants. New EvolinkChatModel enum, EvolinkConfig (extends OpenaiConfig, reads from EVOLINK_ env prefix, validates non-empty api_key), and Evolink class (subclasses OpenAI) defined in evolink.py.
LLM factory selection wiring backend/director/llm/__init__.py	Evolink imported and get_default_llm() extended with an EVOLINK_API_KEY-derived flag; returns Evolink() when DEFAULT_LLM == LLMType.EVOLINK or when the key is present as a fallback.
Tests, env config, and documentation backend/tests/llm/test_evolink.py, backend/.env.sample, backend/README.md, docs/llm/evolink.md, mkdocs.yml	Four tests cover config defaults, env overrides, missing-key ValueError, and get_default_llm() selection. EVOLINK_API_KEY/BASE/CHAT_MODEL added to .env.sample. README updated to mention EvoLink alongside OpenAI. New docs/llm/evolink.md page added and registered in mkdocs.yml.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Poem

A new link evolved in the warren today,
EvoLink hops in with its own API way.
Config defaults set, the key must be there,
Or ValueError jumps out from its lair.
🐇 Now get_default_llm() knows just where to go—
Another provider joins Director's burrow! 🌿

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 28.57% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'feat: add EvoLink LLM provider' directly and clearly describes the main change—adding EvoLink as a new LLM provider to the system.
Linked Issues check	✅ Passed	All objectives from issue #194 are met: EvoLink provider class created, environment variables configured with defaults, DEFAULT_LLM selection logic updated, documentation and tests added.
Out of Scope Changes check	✅ Passed	All changes directly support EvoLink LLM provider integration. The DOWNLOADS_PATH reformatting is a minor formatting adjustment without functional impact.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ 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 and usage tips.}

[EvoLinkAI]

Real EvoLink verification completed from Director provider path:

curl https://direct.evolink.ai/v1/chat/completions with model gpt-5.2 -> HTTP 200
response: Director EvoLink OK

Director Evolink(EvolinkConfig(max_tokens=32, temperature=0)).chat_completions(...)
status= 1
content= Director EvoLink provider OK
model= gpt-5.2
base= https://direct.evolink.ai/v1