ccbot AGENTS.md

AGENTS.md

ccmux — Telegram bot that bridges Telegram Forum topics to Codex sessions via tmux windows. Each topic is bound to one tmux window running one Codex instance.

Tech stack: Python, python-telegram-bot, tmux, uv.

Common Commands

uv run ruff check src/ tests/         # Lint — MUST pass before committing
uv run ruff format src/ tests/        # Format — auto-fix, then verify with --check
uv run pyright src/ccbot/             # Type check — MUST be 0 errors before committing
./scripts/restart.sh                  # Restart the ccbot service after code changes
ccbot hook --install                  # Auto-install Codex SessionStart hook

Core Design Constraints

• 1 Topic = 1 Window = 1 Session — all internal routing keyed by tmux window ID (@0, @12), not window name. Window names kept as display names. Same directory can have multiple windows.
• Topic-only — no backward-compat for non-topic mode. No active_sessions, no /list, no General topic routing.
• No message truncation at parse layer — splitting only at send layer (split_message, 4096 char limit).
• MarkdownV2 only — use safe_reply/safe_edit/safe_send helpers (auto fallback to plain text). Internal queue/UI code calls bot API directly with its own fallback.
• Hook-based session tracking — SessionStart hook writes session_map.json; monitor polls it to detect session changes.
• Message queue per user — FIFO ordering, message merging (3800 char limit), tool_use/tool_result pairing.
• Rate limiting — AIORateLimiter(max_retries=5) on the Application (30/s global). On restart, the global bucket is pre-filled to avoid burst against Telegram's server-side counter.

Code Conventions

• Every .py file starts with a module-level docstring: purpose clear within 10 lines, one-sentence summary first line, then core responsibilities and key components.
• Telegram interaction: prefer inline keyboards over reply keyboards; use edit_message_text for in-place updates; keep callback data under 64 bytes; use answer_callback_query for instant feedback.