• Add Telegram bot integration for task creation from messages
• Create TaskInput database model to store extracted events
• Implement task_input_service for persisting data to database
• Add automatic database table creation on application startup

Add database initialization on startup

• Import Base from database and TaskInput model
• Add startup event handler to create database tables
• Initialize SQLAlchemy metadata on application startup

app/main.py

Create models package initialization

• Create new models package with TaskInput export
• Provide clean module interface for model imports

app/models/init.py

Define TaskInput database model

• Define TaskInput SQLAlchemy ORM model with task_inputs table
• Store source, original text, and extracted event data
• Include optional Telegram chat and message IDs for tracking
• Add created_at timestamp with UTC default

app/models/task_input.py

Description

app/services/telegram_bot.py imports the telegram package, but the project dependency list does
not include python-telegram-bot, so importing/running the bot will fail with
ModuleNotFoundError. This blocks the Telegram integration from working at all.

Code

app/services/telegram_bot.py[R4-6]

Evidence

The new Telegram bot module directly imports telegram/telegram.ext, but the declared runtime
dependencies in pyproject.toml do not include python-telegram-bot, and the lockfile contains no
Telegram package entries, so the import cannot succeed in a clean install.

app/services/telegram_bot.py[1-13]
pyproject.toml[1-13]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`app/services/telegram_bot.py` imports `telegram` / `telegram.ext`, but the repository does not declare a dependency that provides those modules. This will crash at import-time with `ModuleNotFoundError: No module named 'telegram'`.

## Issue Context
- The Telegram bot implementation lives in `app/services/telegram_bot.py`.
- Dependencies are declared in `pyproject.toml`; lockfile should be updated accordingly.

## Fix
1. Add `python-telegram-bot` to `[project].dependencies` in `pyproject.toml`.
2. Re-lock dependencies (e.g., `uv add python-telegram-bot` or equivalent) so `uv.lock` is updated.

## Fix Focus Areas
- pyproject.toml[1-13]
- app/services/telegram_bot.py[1-13]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

Description

.env.example adds TELEGRAM_WEBHOOK_SECRET=some-long-random-secret, which does not look like a
clear placeholder and risks committing a real secret to version control. This violates the
requirement that .env.example contain only non-secret placeholder/example values.

Code

.env.example[R12-13]

Evidence

PR Compliance ID 1 forbids committing real secrets and requires .env.example to contain only
placeholders. The PR adds TELEGRAM_WEBHOOK_SECRET=some-long-random-secret, which is not an
explicit placeholder pattern (e.g., your_webhook_secret_here) and may be interpreted/used as an
actual secret.

CLAUDE.md: Do Not Commit Environment-Specific Secrets; Use Gitignored .env.dev Copied From .env.example
.env.example[12-13]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`.env.example` includes a likely non-placeholder value for `TELEGRAM_WEBHOOK_SECRET`, which may represent a real secret and should not be committed.

## Issue Context
Compliance requires `.env.example` to contain only non-secret placeholder/example values.

## Fix Focus Areas
- .env.example[12-13]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

Description

app/main.py executes Base.metadata.create_all() on every application startup, performing schema
DDL implicitly at runtime. This can cause startup races/locks or inconsistent schema management when
running multiple workers/replicas and makes DB changes harder to control.

Code

app/main.py[R30-33]

Evidence

The new startup handler unconditionally calls Base.metadata.create_all() using the shared engine,
meaning every app process will attempt schema creation on startup.

app/main.py[18-34]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The app runs `Base.metadata.create_all()` during startup. Running DDL automatically at runtime is risky in multi-process/multi-replica deployments and makes schema evolution implicit.

## Issue Context
- `app/main.py` already defines a `lifespan` handler and now also defines a separate startup event that runs DDL.
- The code imports `TaskInput` solely to ensure the model is registered for `create_all()`.

## Fix
Choose one:
1. Preferred: remove the `create_all()` startup hook and manage schema changes via a dedicated migration/init step.
2. If you must keep auto-creation for local/dev: gate it behind an env flag (e.g., `AUTO_CREATE_TABLES=true`) and ensure it runs in a single controlled process (not every worker).

Also consider moving all startup logic into the existing `lifespan` context manager to avoid split initialization.

## Fix Focus Areas
- app/main.py[18-34]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

Description

.env.example adds TELEGRAM_WEBHOOK_SECRET, but the bot code runs in polling mode and no
application code reads this secret. This is misleading configuration that will drift or confuse
deploys.

Code

.env.example[R12-13]

Evidence

The example env file introduces the secret, but the Telegram bot uses run_polling() and no Python
files reference TELEGRAM_WEBHOOK_SECRET, so the configuration is currently unused.

.env.example[10-13]
app/services/telegram_bot.py[70-83]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`TELEGRAM_WEBHOOK_SECRET` is documented in `.env.example` but is not used anywhere in code; the Telegram integration uses polling.

## Issue Context
- `app/services/telegram_bot.py` starts polling via `run_polling()` and does not implement webhook request validation.

## Fix
Either:
- Remove `TELEGRAM_WEBHOOK_SECRET` from `.env.example` until webhook support is implemented, or
- Implement webhook mode + verification and then keep the env var.

## Fix Focus Areas
- .env.example[11-13]
- app/services/telegram_bot.py[70-83]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools