Loading repository data…
sen-rce / repository
Get a Telegram message the moment your Claude Code usage limits reset. Pure-Python, zero-dependency bot that an AI agent builds and sets up for you from a single prompt.
A transparent discovery signal based on current public GitHub metadata.
This score does not audit code, security, maintainers, documentation quality, or suitability. Verify the repository and its current documentation before adoption.
Get a Telegram message the moment your Claude Code limits reset — stop refreshing
/usageby hand. You don't build it: hand one file to Claude Code and the agent builds and sets it all up for you.
pip install, no dependencies.seven_day_opus, …).git clone https://github.com/<you>/claude-limits-telegram-bot.git
cd claude-limits-telegram-bot
# open this folder in Claude Code, then send:
# follow AGENT_BUILD.md
That's it. The agent reads the spec, checks the code, and walks you through the whole setup interactively — then shows you a real reset notification to prove it works.
The agent handles the code, .env, statusLine registration, autostart, tests,
and the demo. From you it needs just three small things — it will ask at the
right moment:
/newbot) and paste
the token to the agent when asked. (This is the only secret you provide.)Total time: ~5 minutes. After setup you only need to restart Claude Code once so the limit feed starts flowing.
While you use Claude Code, its official statusLine feed reports your limit
usage for free (no token, no rate limit). A tiny background daemon watches that
feed and pushes you a Telegram message when a window resets. When you're idle, it
lazily falls back to the undocumented /api/oauth/usage endpoint just long
enough to catch the reset — never on a timer.
The spec the agent follows: docs/SPEC.md ·
docs/TASKS.md. A finished reference implementation
(bot.py, usage.py, ) is included at the repo root.
Selected from shared topics, language and repository description—not editorial ratings.
WHOOPsClaw /
Self-hosted AI news digest for Telegram. Describe your interests in plain text, get curated daily briefings. Supports 6 source types (RSS, Telegram, Reddit, HN, YouTube, Search) and 8 LLM providers (Groq free by default, OpenAI, Claude, and more). One docker compose up, 5-question setup, done.
46/100 healthstatusline_feed.pyYou can skip the agent and run the reference implementation directly.
python3 — already on macOS).Open @BotFather, /newbot, copy the token — that's
TG_BOT_TOKEN.
Send your bot a message, open
https://api.telegram.org/bot<YOUR_TOKEN>/getUpdates, find
"chat":{"id":NNNNNNN} — that's TG_CHAT_ID.
Add to ~/.claude/settings.json (absolute path, quoted if it has spaces):
{
"statusLine": {
"type": "command",
"command": "python3 \"/absolute/path/to/claude-limits-telegram-bot/statusline_feed.py\""
}
}
⚠️ Overwrites an existing custom statusLine. Restart Claude Code to
activate; after your next turn a usage_feed.json appears next to the script.
cp .env.example .env
chmod 600 .env
# fill in TG_BOT_TOKEN and TG_CHAT_ID
python3 bot.py
Send /start, /status, /help to your bot.
gui/$UID (Keychain access for the fallback token).All optional except the two tokens; set in .env or the environment
(environment wins). Full list in .env.example.
| Variable | Purpose | Default |
|---|---|---|
TG_BOT_TOKEN | Telegram bot token | required |
TG_CHAT_ID | Your chat ID | required |
USAGE_FEED_PATH | statusLine feed file | ./usage_feed.json |
STATE_FILE | dedup / offset state | ./bot_state.json |
FEED_STALE_SEC | feed age before it's "stale" | 600 |
RESET_BUFFER_SEC | wait after reset before verifying | 30 |
POLL_BACKOFF_SEC | backoff on 429 without Retry-After | 600 |
WARN_THRESHOLD | optional %-used pre-warning | off |
TIMEZONE | tz for message times (e.g. Europe/London) | system |
LOG_LEVEL | INFO / DEBUG / ... | INFO |
/status says "no fresh data" — do a turn in Claude Code so the feed
populates, and make sure you restarted Claude Code after registering statusLine.gui/$UID.claude and /login again. statusLine keeps
working meanwhile./api/oauth/usage is undocumented and aggressively rate-limited. It's only
used lazily (never on a timer), and the anthropic-beta header is versioned — it
may break someday. If it does, the bot degrades gracefully to the statusLine feed.
python3 -m unittest discover -s tests
MIT — see LICENSE.