1# zot
2
3yet another coding agent harness, lightweight and written (vibe-slopped) in go.
4
5- one static binary.
6- two providers atm (anthropic, openai/codex).
7- four tools (read, write, edit, bash).
8- three run modes (interactive tui, print, json).
9- built-in telegram bot.
10- extensions in any language via subprocess + json-rpc — none installed by default; opt in with `zot ext install` or `zot --ext`. see [docs/extensions.md](docs/extensions.md).
11- reusable instructions via `SKILL.md` files; see [docs/skills.md](docs/skills.md).
12- no community atm.
13
14## install
15
16### one-liner (macos + linux)
17
18```bash
19curl -fsSL https://raw.githubusercontent.com/patriceckhart/zot/main/install.sh | bash
20```
21
22detects your os/arch, downloads the latest release from github, verifies the sha256 against the release's `checksums.txt`, extracts the binary, and drops it in `/usr/local/bin`, `~/.local/bin`, or `~/bin` — whichever is writable first. pass a version or prefix to pin:
23
24```bash
25curl -fsSL https://raw.githubusercontent.com/patriceckhart/zot/main/install.sh | bash -s -- v0.0.1 ~/bin
26```
27
28### one-liner (windows, powershell)
29
30```powershell
31iwr -useb https://raw.githubusercontent.com/patriceckhart/zot/main/install.ps1 | iex
32```
33
34drops `zot.exe` into `$HOME\bin` and adds it to the user PATH if missing. open a fresh terminal afterwards.
35
36### homebrew (macos + linux)
37
38```bash
39brew install patriceckhart/tap/zot
40```
41
42the tap lives at `patriceckhart/homebrew-tap`.
43
44### go install
45
46```bash
47go install github.com/patriceckhart/zot/cmd/zot@latest
48```
49
50### from source
51
52```bash
53git clone https://github.com/patriceckhart/zot
54cd zot
55make build        # produces ./bin/zot
56make install      # into $GOPATH/bin
57```
58
59### prebuilt binaries
60
61every release on the [releases page](https://github.com/patriceckhart/zot/releases) ships archives for linux, macos, and windows on amd64 + arm64 (except windows/arm64), plus a `checksums.txt` file. download, verify, `chmod +x`, and drop on your `$PATH`.
62
63## authenticate
64
65the easiest way is to just run `zot` and type `/login`. the tui opens even without credentials and walks you through a browser-based login flow.
66
67### credential lookup order
68
691. `--api-key` flag
702. `ANTHROPIC_API_KEY` / `OPENAI_API_KEY` env var
713. `$ZOT_HOME/auth.json` (api key or oauth token; mode 0600)
72
73`$ZOT_HOME` defaults to:
74- macOS: `~/Library/Application Support/zot`
75- linux: `$XDG_STATE_HOME/zot` or `~/.local/state/zot`
76- windows: `%LOCALAPPDATA%\zot`
77
78### `/login` flow
79
80run `zot` and type `/login`. pick one of two methods:
81
82- **api key** — a small local web server starts on `127.0.0.1:<free-port>`, your browser opens a form, you paste your `sk-ant-...` or `sk-...` key. zot probes the provider once and saves it to `auth.json` if accepted.
83- **subscription** — use your claude pro/max or chatgpt plus/pro subscription. the oauth flow pins the callback to a fixed port per provider (`localhost:53692` for anthropic, `localhost:1455` for openai) because those are the only ports their auth servers will redirect to.
84  - anthropic uses the claude code oauth flow; messages go to `api.anthropic.com` with a bearer token and the claude-code identity headers.
85  - openai uses the codex cli oauth flow; messages go to `chatgpt.com/backend-api/codex/responses` with the `chatgpt-account-id` extracted from the returned id_token.
86
87> **note on subscription login**: the oauth client ids used are the ones published in anthropic's claude code cli and openai's codex cli. reusing them from a third-party tool is against their terms of service and may be revoked at any time. use it at your own risk; the api-key flow is the safe default.
88
89### token refresh
90
91oauth access tokens are short-lived (anthropic ~8h, openai ~30d). zot refreshes them automatically:
92
93- at every credential lookup, zot checks the stored `expiry` and — if past it (with a 60s safety margin) — hits the provider's `oauth/token` endpoint with the stored `refresh_token`, persists the new `access_token` + `refresh_token` + `expiry` back to `auth.json`, and hands the fresh token to the client.
94- the telegram bridge additionally refreshes once per turn so a bot that runs for days keeps working without manual intervention.
95- if the refresh itself fails (the `refresh_token` was revoked, or the account was logged out everywhere), the error bubbles up to the caller: the tui shows it in the status line, the bot replies with it in your dm. run `/login` to get a fresh token pair.
96
97all data lives under `$ZOT_HOME`:
98
99```
100$ZOT_HOME/
101├── config.json         # last-used provider/model/theme, saved automatically
102├── auth.json           # api keys and oauth tokens (mode 0600)
103├── sessions/           # jsonl transcripts, one dir per cwd
104├── models-cache.json   # live /v1/models discovery cache (6h ttl)
105├── SYSTEM.md           # optional: replaces the default system prompt
106└── logs/               # app log files
107```
108
109drop a `SYSTEM.md` in `$ZOT_HOME` to replace the built-in identity + guidelines for every run. `--system-prompt` still wins per-invocation. delete the file to revert to the default.
110
111## usage
112
113```bash
114zot                              # interactive tui
115zot "fix the failing test"       # tui, pre-filled prompt
116zot -p "list all go files"       # print final text, exit
117zot --json "refactor main.go"    # newline-delimited json events, exit
118zot --continue                   # resume the most recent session for this cwd
119zot --resume                     # pick a session to resume
120zot --list-models                # show supported models
121zot --help
122```
123
124## flags
125
126| flag | description |
127|---|---|
128| `--provider anthropic\|openai` | pick the provider |
129| `--model <id>` | pick the model (see `--list-models`) |
130| `--api-key <key>` | override api key |
131| `--base-url <url>` | override provider base url (tests / self-hosted) |
132| `--system-prompt <text>` | replace the default system prompt for this run (also overrides `$ZOT_HOME/SYSTEM.md`) |
133| `--append-system-prompt <text>` | append text to the system prompt (repeatable) |
134| `--reasoning low\|medium\|high` | enable reasoning on supported models |
135| `-c`, `--continue` | resume the latest session for this cwd |
136| `-r`, `--resume` | pick a session to resume |
137| `--session <path>` | resume a specific session file |
138| `--no-session` | don't read or write session files |
139| `--cwd <path>` | use `<path>` as the working directory |
140| `--no-tools` | disable all tools |
141| `--tools <csv>` | only enable the listed tools |
142| `--max-steps <n>` | cap agent loop iterations (default 50) |
143
144## tools
145
146- `read` — read text files (or inline images: png / jpg / gif / webp)
147- `write` — create or overwrite files, making parent directories as needed
148- `edit` — one or more exact-match replacements in an existing file
149- `bash` — run a shell command in the session cwd, with merged stdout/stderr and a timeout
150
151when the sandbox is on (see `/lock`), all four tools refuse paths outside the session cwd.
152
153## modes
154
155- **interactive** (default): chat tui with streaming output, spinner, cost meter, slash commands.
156- **print**: `zot -p "prompt"` runs the agent to completion and writes only the final assistant text to stdout.
157- **json**: `zot --json "prompt"` emits one json object per agent event to stdout, newline-delimited. the schema is documented in `instructions.md` §8.
158- **rpc**: `zot rpc` runs as a long-lived child process; commands in on stdin, events + responses out on stdout, both as ndjson. designed for embedding zot in third-party apps written in any language. see `docs/rpc.md` for the wire schema and `examples/rpc/{python,node,shell,go}` for working clients.
159
160## embedding
161
162two ways to drive zot from another program:
163
164- **go in-process**: import `github.com/patriceckhart/zot/pkg/zotcore`. one `Runtime` per project; `Prompt(ctx, text, images)` returns a channel of `Event`. small example in `examples/sdk/`.
165- **any language out-of-process**: spawn `zot rpc` as a subprocess and exchange newline-delimited json over its stdin/stdout. wire format and event schema in [docs/rpc.md](docs/rpc.md). reference clients live under `examples/rpc/`.
166
167both interfaces share the same event schema, so transcripts captured by one can be replayed through the other.
168
169## slash commands
170
171type `/` in the tui to open the autocomplete popup. available commands:
172
173| command | description |
174|---|---|
175| `/help` | show key bindings and commands |
176| `/login` | log in via api key or subscription (opens a dialog) |
177| `/logout [provider]` | clear credentials for `anthropic`, `openai`, or all when omitted |
178| `/model` | pick a model from a list (or `/model <id>` to set directly) |
179| `/sessions` | resume a previous session for this directory |
180| `/jump` | scroll the chat to a previous turn (or `/jump <text>` to filter) |
181| `/btw` | side-chat with full context that doesn't add to the main thread |
182| `/compact` | summarize the transcript into one message to free up context |
183| `/lock` | confine tools to the current directory |
184| `/unlock` | allow tools to touch paths outside again |
185| `/clear` | clear the chat transcript |
186| `/exit` | exit zot |
187
188### `/sessions`
189
190shows previous sessions for the current working directory, newest first, with timestamp, model, message count, cost, and the first user prompt. pick one with `↑/↓`, `enter` to resume, `esc` to cancel. zot swaps the current session file for the selected one and replays the full transcript (including tool calls) into the agent. sessions remember the model they ended on, so resuming picks up on that exact model even if your global default changed.
191
192### `/jump`
193
194opens a turn picker for the current session — one row per user prompt, each showing the turn number, how many tools that turn invoked, and the first line of the prompt. `↑/↓` to pick, `enter` to jump, `esc` to cancel. any printable rune while the picker is open extends a filter; backspace narrows it back. `/jump <text>` pre-applies the filter; if exactly one turn matches, zot jumps straight there without showing the picker.
195
196jumping is non-destructive — the transcript is untouched, the viewport just scrolls so the chosen turn is at the top. a muted line at the top of the chat reads `↑ viewing turn N of M · pgdn to catch up`; scroll back to the bottom with `pgdn` (or keep scrolling with the arrow keys) and the indicator goes away.
197
198### `/btw`
199
200opens a side-chat overlay with the full main session as frozen context, so you can ask quick clarifying questions ("does asyncio.gather() catch exceptions?", "btw the bundle budget is 10MB", "what's the default fetch timeout?") without bloating the main thread.
201
202each question fires a one-off model call against `system + main transcript + side-chat history so far`. responses render in the overlay and stay there. when you press `esc` to close, **nothing** has been added to the main session and subsequent main-thread turns don't re-read any of the side-chat exchanges — keeping the running context window lean.
203
204```
205/btw                              # open the overlay, type questions interactively
206/btw does PUT replace the whole resource?
207```
208
209inside the overlay: `enter` sends, `esc` cancels an in-flight call (or closes the overlay if idle), `ctrl+c` closes immediately. side-chat exchanges never touch the transcript and aren't persisted to the session file.
210
211### `/compact`
212
213sends the current transcript through the model with a structured summarization prompt. the returned summary replaces the transcript as one synthetic user message, with the last few exchanges kept verbatim for continuity. status bar's `ctx N/M (P%)` meter resets. use it when the context meter creeps past ~80%.
214
215zot also auto-compacts in the background: after any turn that leaves context usage ≥ **85%** of the model's window, the agent kicks off a condense pass on its own. you'll see `condensing history… (esc to cancel)` above the status bar and an `(auto)` tag next to the context percentage; esc aborts it without touching the transcript.
216
217### `/lock`
218
219enforces a sandbox rooted at the cwd shown in the status bar. `read` / `write` / `edit` resolve their target path (including through symlinks) and refuse anything outside the sandbox. `bash` refuses obvious escape patterns: `sudo`, `rm -rf /`, leading `cd /` / `cd ..` / `cd ~`, `chmod -R`, `dd of=/`, etc. status bar shows `· locked · ~/your/cwd` while active.
220
221this is a guardrail against accidents, not a hard security boundary. if you need real isolation, run zot under docker or a proper sandbox.
222
223## sessions
224
225every interactive or print/json run (unless `--no-session`) writes a jsonl transcript under `$ZOT_HOME/sessions/<cwd-hash>/`. resume any of them with `--continue`, `--resume`, `--session <path>`, or interactively via `/sessions` inside the tui.
226
227## models
228
229`--list-models` or the `/model` picker shows the full catalog. three sources:
230
231- **catalog** — models baked into zot, always available
232- **live** — ids discovered from `GET /v1/models` using your stored api key (cached for 6h in `$ZOT_HOME/models-cache.json`, refreshed in the background on startup)
233- **speculative** — ids that appear in the upstream generator but aren't live on the public api yet; they'll 404 today and start working the moment the provider ships them
234
235the context meter in the status line (`ctx N/M (P%)`) uses the model's advertised context window to show how much of it your last turn consumed.
236
237## inline images
238
239when a tool returns an image (e.g. `read` on a png), zot renders it inline on terminals that support it: **iterm2**, **wezterm**, **kitty**, **ghostty**. on other terminals you see a text placeholder with mime type, pixel dimensions, and byte size. control with the `ZOT_INLINE_IMAGES` env var:
240
241| value | effect |
242|---|---|
243| unset (default) | auto-detect based on `TERM_PROGRAM` |
244| `iterm` / `iterm2` | force iterm2 osc 1337 protocol |
245| `kitty` | force kitty graphics protocol |
246| `off` / `none` | always use the text placeholder |
247
248frames containing images are full-repainted (no differential diff) to prevent stale image pixels from lingering through scroll. that costs one terminal flash per image-containing frame; set `ZOT_INLINE_IMAGES=off` if that bothers you.
249
250## queued messages
251
252you can keep typing while the agent is working. pressing enter during a turn queues the message instead of interrupting: it shows up above the status bar as `▸ sliding in: <text>` and is delivered as the next user turn the moment the current one finishes. queue as many as you want; they run in order. esc / ctrl+c cancels the active turn and drops the queue so a runaway turn doesn't flood you with stale follow-ups.
253
254slash commands also work while the agent is busy. read-only ones (`/help`, `/jump`, `/btw`, `/sessions`, `/lock`, `/unlock`, `/exit`) take effect immediately. destructive ones (`/clear`, `/compact`, `/login`, `/logout`, `/model`) cancel the active turn first and then run.
255
256## keys (interactive mode)
257
258### input
259
260| key | action |
261|---|---|
262| `enter` | submit (queued if the agent is busy) |
263| `alt+enter` | newline |
264| `tab` | complete the selected slash command |
265| `esc` | cancel the current turn (while busy); clear input (while idle) |
266| `ctrl+c` | clear the input + queue (or cancel the current turn). press again within 2s to exit. |
267| `ctrl+d` | exit on empty input |
268| `ctrl+l` | redraw the screen |
269| `ctrl+o` | expand / collapse long tool results |
270
271### editor line navigation
272
273| key | action |
274|---|---|
275| `ctrl+a` / `ctrl+e` | jump to start / end of line |
276| `alt+←` / `alt+→` | jump one word back / forward |
277| `ctrl+u` / `ctrl+k` | delete to start / end of line |
278| `ctrl+w` · `alt+backspace` | delete the previous word |
279| `up` / `down` (editor non-empty) | cycle through prompt history |
280
281### chat scroll
282
283| key | action |
284|---|---|
285| `pgup` / `pgdn` | scroll one page up / down |
286| `up` / `down` (editor empty) | scroll three lines up / down — this is how the mouse wheel reaches the scroll logic on most terminals |
287
288## telegram bot (bridge)
289
290zot can run as a telegram bot so you can dm it from your phone. it's a built-in subcommand, not a plugin:
291
292```bash
293zot telegram-bot setup     # paste a BotFather token, verify, save
294zot telegram-bot run       # foreground: long-poll in this terminal (ctrl+c to stop)
295zot telegram-bot start     # background: detach and return immediately
296zot telegram-bot stop      # sigterm the background bot (sigkill after 5s)
297zot telegram-bot logs -f   # tail $ZOT_HOME/logs/bot.log (omit -f to just cat)
298zot telegram-bot status    # config (token masked) + running/stopped
299zot telegram-bot reset     # forget the token + paired user
300# short alias: `zot tg ...` is accepted for every subcommand
301```
302
303the background flavor writes the child's pid to `$ZOT_HOME/bot.pid` and redirects stdout+stderr to `$ZOT_HOME/logs/bot.log`. `zot telegram-bot stop` reads that pid, sends sigterm, waits up to five seconds, then escalates to sigkill if the child is still alive. running two instances at once is refused at startup.
304
305> **use the installed binary for `start`.** `go run ./cmd/zot telegram-bot start` won't work — `go run` builds a binary in a temp directory and deletes it when it exits, which kills the detached child. run `make install` (or `go build`) first and invoke the installed binary.
306
307setup flow:
308
3091. talk to [@BotFather](https://t.me/BotFather) on telegram, run `/newbot`, copy the token it gives you.
3102. run `zot telegram-bot setup` and paste the token when prompted.
3113. run `zot telegram-bot run` in the directory you want the agent to operate in.
3124. open your bot on telegram, send `/start`. the first user to do this claims the bridge (stored as `allowed_user_id`); every other user is rejected.
313
314from then on, any dm you send is forwarded to the agent as a user prompt. attached photos or image/* documents are downloaded and passed to vision-capable models. in-bot telegram commands: `/help`, `/status`, `/stop` (cancel the current turn). config lives in `$ZOT_HOME/bot.json` (mode 0600).
315
316bot mode respects the usual zot flags — `--provider`, `--model`, `--cwd`, `--reasoning`, `--continue`, `--no-session`, `--no-tools`, etc. run `zot tg run -c --model claude-opus-4-1` to resume the latest session on opus, for example.
317
318## development
319
320```bash
321make build     # build ./bin/zot
322make test      # go test -race ./...
323make lint      # go vet + gofmt check
324make fmt       # gofmt -w .
325make release   # cross-compile linux/darwin/windows × amd64/arm64
326```
327
328source layout:
329
330```
331cmd/zot/                      main()
332internal/agent/               cli wiring, arg parsing, system prompt, config
333internal/agent/modes/         interactive tui, print, json, dialogs
334internal/agent/tools/         read, write, edit, bash, sandbox
335internal/auth/                credential store, api-key probe, oauth, login server
336internal/core/                agent loop, sessions, cost tracking
337internal/provider/            anthropic + openai streaming clients, model catalog
338internal/tui/                 terminal raw-mode, input parser, editor, renderer, markdown, view
339```
340
341## license
342
343MIT