# clisbot v0.1.39

Released: 2026-04-18

This is the first large release note for the current `clisbot` shape.

It covers the biggest user-facing work shipped between 2026-04-12 and 2026-04-18.

I now use `clisbot` to develop `clisbot` itself, turning it into an invaluable product feedback loop.

## In One Line

`clisbot` now feels much closer to a real Slack and Telegram assistant instead of a thin CLI bridge: it is easier to start, much easier to read inside the native channel surface, safer to expose, better at long-running work, and more usable day to day.

## What Feels Different

### 1. Starting the first bot is much easier

- First-run setup is now much more direct, with a clearer `clisbot start --cli ... --bot-type ...` path.
- Slack and Telegram setup guides are now much stronger and much easier to follow.
- Persistent credentials and runtime-only credentials are easier to understand, so trying the bot is faster without locking you into a final setup too early.
- `CLISBOT_HOME` support makes it much easier to run isolated environments for dev, test, or a second bot instance.

### 2. The bot feels more native inside chat

- Slack and Telegram are treated much more like real product surfaces, not just plain text outputs.
- Channel-native rendering is much better and much easier to read, including native Slack table rendering and stronger Slack or Telegram formatting control.
- Routing, thread or topic continuity, and surface-specific reply behavior are clearer and more predictable.
- Recent conversation replay helps the bot pick up context more naturally in active conversations.

### 3. Long-running work is easier to trust

- Streaming previews, running indicators, and final message settlement are much more truthful.
- Tmux startup, submit, attach, detach, and recovery flows were hardened heavily, so long jobs are less fragile.
- The bot is better at surviving slow starts, runner hiccups, and mid-run interruptions without leaving users confused.
- Runtime monitor backoff and owner alerts make failures more visible instead of silently messy.

### 4. You can now automate repeated work

- `/loop` support landed, so an agent can now run recurring work on a schedule.
- Loop control CLI support landed too, so operators can inspect and cancel managed loops cleanly.
- Queue and loop start notifications make background work more visible from the chat surface.
- `/nudge` support gives a lightweight way to push a stuck run forward without restarting the whole flow.

### 5. Shared use is much safer and clearer

- DMs default to pairing-oriented access instead of silent open access.
- App auth and routed auth are much stronger, including clearer roles, permissions, and enforcement.
- First-owner auto-claim makes the first secure bootstrap path much less awkward.
- Slack shared-channel behavior is safer by default, including stronger mention requirements and better guidance when a route is missing.

### 6. Operators can see and debug more

- Status and runtime summaries are much richer and more honest.
- Runner debug surfaces are much better, including `runner watch --latest` and stronger recent-session visibility.
- Setup, workflow, and operator docs were reshaped around the bot-first mental model, so the whole product is easier to reason about.

### 7. More runner choice is available

- Gemini CLI runner support landed, so `clisbot` now works across Codex, Claude, and Gemini families with a more unified operator story.

## Biggest Wins By Persona

### If you are just chatting with the bot

- replies feel more natural and less confusing
- long tasks are easier to follow
- the bot remembers nearby context better
- Slack and Telegram output is much easier to read

### If you are setting the bot up

- first-run is much easier
- bot credentials are easier to reason about
- setup docs are much clearer
- separate dev and main environments are easier to keep clean

### If you run `clisbot` for a team

- access control is much stronger
- shared-channel defaults are safer
- failures are more visible
- recurring work is now a real product path

## Highlights

- One-command first run is much more realistic now.
- Slack and Telegram feel more native instead of feeling like terminal mirrors, and the output is much easier to read at a glance.
- Long-running runs are far more resilient and reviewable.
- Scheduled loop automation is now part of the product.
- Auth, pairing, and first-owner setup are much closer to something teams can trust.
- Operator status and runner debugging are much stronger.

## Important Upgrade Note

- `v0.1.39` includes breaking changes in the official config shape and in the main CLI command surface.
- If you already installed `clisbot` before this release, ask Codex or Claude inside the repo to update your current config into the `v0.1.39` shape before you upgrade.
- The docs in this release now assume the current official config and command model.

## Upgrade Path

Upgrade itself is still very simple:

```bash
clisbot stop
npm install -g clisbot
clisbot start
clisbot --version
```

Recommended path for existing installs:

1. Update your config shape first.
2. Run the upgrade commands above.
3. Confirm `clisbot --version` shows `0.1.39`.
4. Start with one real bot and one real route first, then expand.

## Recommended Docs After This Release

- [CHANGELOG.md](../../CHANGELOG.md)
- [Release Notes Index](README.md)
- [User Guide](../user-guide/README.md)
- [Bots And Credentials](../user-guide/bots-and-credentials.md)
- [CLI Commands](../user-guide/cli-commands.md)
- [Telegram Bot Setup](../user-guide/telegram-setup.md)
- [Slack App Setup](../user-guide/slack-setup.md)