|
|
13e058f22c
|
feat: add openai-compatible provider for custom AI endpoints
Add AI_PROVIDER=openai-compatible mode that reuses OpenAISynthesisProvider
with a validated custom base_url, allowing any OpenAI-compatible API
(OpenRouter, Ollama, LiteLLM proxy, etc.) without new code.
Configuration:
- AISettings.provider now accepts openai-compatible
- New AISettings.allow_insecure_http: bool = False (HTTP opt-in)
- .env.example: commented examples for OpenRouter (HTTPS) and Ollama (HTTP)
Factory validation (_validate_openai_compatible_config):
- base_url and model required, api_key required (MVP)
- HTTPS enforced unless allow_insecure_http=true
- Credentials in URL rejected, sensitive query params rejected
(including valueless params via keep_blank_values=True)
- Malformed URLs and missing hostname rejected (ValueError caught)
- No /v1 manipulation; degraded to None + warning on invalid config
- redact_url() used for all URL warnings
Tests: 13 new factory tests in test_synthesis.py covering routing,
URL validation, HTTP policy, credentials, sentinel non-leak, no-network.
Coverage: 91.57% (synthesis module).
Docs: GUIDE_DEV_PYTHON.md §9.5 updated with 3-provider table, validation
rules, and synchronized code example.
mypy override for openai.* (follow_imports=skip) to work around
mypy 2.3.1 internal error in pre-commit's isolated environment.
Co-authored-by: opencode/coder anthropic.claude-sonnet-4-5 <anthropic.claude-sonnet-4-5@agents.invalid>
Co-authored-by: opencode/test-engineer anthropic.claude-sonnet-4-5 <anthropic.claude-sonnet-4-5@agents.invalid>
Co-authored-by: opencode/tech-writer anthropic.claude-sonnet-4-5 <anthropic.claude-sonnet-4-5@agents.invalid>
|
2026-09-07 19:44:40 +02:00 |
|
|
|
19cbf8f13f
|
fix(M9): corrections d'audit FIXME_M9 — secrets, messages, validation, tests
Cinq corrections de l'audit FIXME_M9 :
- redact_secrets() étendue avec extra_secrets pour masquer les clés brutes ;
providers stockent SecretStr jusqu'à l'appel SDK.
- _build_prompt() inclut le contenu des messages (tronqué à 500 car.) ;
prompt système renforcé contre l'injection.
- _validate_output() supprime les emojis et rejette titre/liste/HTML → None.
- Tests litellm utilisent importorskip + LITELLM_LOCAL_MODEL_COST_MAP=true.
- .env.example désactive l'IA par défaut (AI_ENABLED=false).
- Documentation §9.2-§9.5 alignée avec l'implémentation (SDK openai, SecretStr,
factory réelle, validation sortie, politique hors réseau).
Co-authored-by: opencode/coder <coder@agents.invalid>
Co-authored-by: opencode/test-engineer <test-engineer@agents.invalid>
Co-authored-by: opencode/tech-writer <tech-writer@agents.invalid>
|
2026-09-07 19:01:25 +02:00 |
|
|
|
92833060e2
|
feat(M9): synthèse IA — protocole, providers OpenAI/litellm, factory, tests
Synthèse optionnelle via SDK openai (client injectable, prompt système
FR, max 800 car., timeout 30 s, temp 0.3). Mode dégradé strict :
generate() ne lève jamais, retourne None si clé absente/timeout/erreur.
Provider litellm optionnel (extra ai-litellm) réutilisant le prompt
OpenAI. Factory get_synthesis_provider() selon AISettings. 23 tests
sans réseau, couverture synthesis/ 93%.
Co-authored-by: opencode/coder <coder@agents.invalid>
Co-authored-by: opencode/test-engineer <test-engineer@agents.invalid>
|
2026-09-07 17:07:05 +02:00 |
|
