Implement Python email agent with multi-model AI support

Adds a complete email-based registration agent for Spielgruppe Pumuckl
based on the OpenSpec define-project-scope specifications.

Architecture
- Channel-agnostic EmailAgent core — no email-specific code in business logic
- Pluggable AI provider layer: Anthropic (Claude) and OpenAI (GPT) supported
  via a shared LLMProvider interface; switch with AI_PROVIDER env var
- IMAP polling for inbound emails with thread-tracking via email headers
  (Message-ID / In-Reply-To / References)
- SMTP for outbound replies and admin notifications
- File-based JSON storage for conversation state and completed registrations
- Admin-editable knowledge-base loaded from markdown files at startup

Key files
  src/config.py                  — env-var configuration
  src/providers/base.py          — abstract LLMProvider
  src/providers/anthropic_provider.py — Claude backend
  src/providers/openai_provider.py    — OpenAI backend
  src/agent/core.py              — EmailAgent orchestrator
  src/agent/prompts.py           — system prompt builder (KB + registration state)
  src/models/registration.py     — RegistrationData matching the JSON schema
  src/models/conversation.py     — ConversationState persisted per thread
  src/channels/email_channel.py  — IMAP/SMTP I/O + quoted-text stripping
  src/storage/json_store.py      — conversation & registration persistence
  src/notifications/notifier.py  — admin notification routing by playgroup type
  main.py                        — polling entry point
  requirements.txt               — anthropic, openai, python-dotenv, jsonschema
  .env.example                   — configuration template

https://claude.ai/code/session_01HaUFs7SaLD5SoiuGCY27Tw
This commit is contained in:
Claude
2026-02-20 20:17:00 +00:00
parent 4d44d4ee58
commit b82ff27efd
24 changed files with 1728 additions and 0 deletions
+72
View File
@@ -0,0 +1,72 @@
"""Configuration loaded from environment variables."""
import os
from dataclasses import dataclass, field
from pathlib import Path
try:
from dotenv import load_dotenv
load_dotenv()
except ImportError:
pass # python-dotenv is optional
@dataclass
class Config:
# AI Provider
ai_provider: str = "anthropic" # "anthropic" or "openai"
ai_model: str = ""
anthropic_api_key: str = ""
openai_api_key: str = ""
# Email — IMAP (receiving)
imap_host: str = ""
imap_port: int = 993
imap_username: str = ""
imap_password: str = ""
imap_use_ssl: bool = True
# Email — SMTP (sending)
smtp_host: str = ""
smtp_port: int = 587
smtp_use_tls: bool = True
# Registration email address shown to parents
registration_email: str = ""
# Storage
data_dir: Path = field(default_factory=lambda: Path("data"))
knowledge_base_dir: Path = field(
default_factory=lambda: Path(
"openspec/changes/define-project-scope/content/knowledge-base"
)
)
# Polling interval in seconds
poll_interval: int = 60
@classmethod
def from_env(cls) -> "Config":
return cls(
ai_provider=os.getenv("AI_PROVIDER", "anthropic"),
ai_model=os.getenv("AI_MODEL", ""),
anthropic_api_key=os.getenv("ANTHROPIC_API_KEY", ""),
openai_api_key=os.getenv("OPENAI_API_KEY", ""),
imap_host=os.getenv("IMAP_HOST", ""),
imap_port=int(os.getenv("IMAP_PORT", "993")),
imap_username=os.getenv("IMAP_USERNAME", ""),
imap_password=os.getenv("IMAP_PASSWORD", ""),
imap_use_ssl=os.getenv("IMAP_USE_SSL", "true").lower() == "true",
smtp_host=os.getenv("SMTP_HOST", ""),
smtp_port=int(os.getenv("SMTP_PORT", "587")),
smtp_use_tls=os.getenv("SMTP_USE_TLS", "true").lower() == "true",
registration_email=os.getenv("REGISTRATION_EMAIL", ""),
data_dir=Path(os.getenv("DATA_DIR", "data")),
knowledge_base_dir=Path(
os.getenv(
"KNOWLEDGE_BASE_DIR",
"openspec/changes/define-project-scope/content/knowledge-base",
)
),
poll_interval=int(os.getenv("POLL_INTERVAL", "60")),
)