Telegram bot
After this page your team’s Telegram chat receives deploy-regression and CU-headroom pushes,
and anyone in it can ask /why <sig> and get a decoded answer.
Connect a chat
Section titled “Connect a chat”The quickstart path: npx @wiretap-labs/cli init prints a
https://t.me/<bot>?start=<token> deep link that binds the chat to your project. To connect a
group chat instead, add the bot to the group and paste the same start token with
/connect <token>.
Commands (pull)
Section titled “Commands (pull)”| Command | Answer |
|---|---|
/why <sig> |
Decoded failure story for a signature: IDL error name, instruction, CU, payer bot score, issue history |
/status |
Program health right now: landing rate vs baseline, open regressions, provider split |
/top [7d] |
Top issues, bot traffic segmented out |
/mute <issue> |
Stop pushes for one issue fingerprint |
/help |
The list above, in chat |
Example /why reply:
✗ swap failed — SlippageExceeded (6001)program DemoAMM… · slot 312 441 907 (devnet)ix swap (top level)cu 48 211 / 200 000 requestedpayer 9xQe…3fKm (bot score 0.02 — real user)Seen 14× this week · issue v1:9c41ae…Pushes
Section titled “Pushes”Two alert classes are pushed without being asked (both deduplicated and rate-limited):
-
Deploy regression — after a detected deploy, error rates are diffed pre/post deploy slot; a significant post-deploy degradation pushes within minutes:
⚠ regression after deploy @ slot 312 440 001InsufficientLiquidity (6004) on swap:0.4% → 7.9% of real-user tx (last 30 min vs 24h pre-deploy) -
CU headroom — when an instruction’s p95 compute usage trends toward its requested budget:
⚠ cu headroom: open_position p95 at 187 300 / 200 000 (6% headroom, shrinking)
Real token
Section titled “Real token”The mock transport upgrades to real Telegram in ~5 minutes:
- Message @BotFather →
/newbot, pick a name and a username ending inbot. - Put the token in
.env:TELEGRAM_BOT_TOKEN=123456:ABC… - Restart the bot app.
wiretap doctorshould reporttelegram: real transport.