pcode in a dark terminal pcode in a light terminal # pcode pcode is a terminal-native coding agent that lets you change what's visible in your scrollback, even after the work is done. Expand commands, output, and diffs when you need the details; hide them when you want to read the conversation. The same history, with the detail you choose. Long-running commands continue in the background and wake the agent when they finish. Run sessions in separate git worktrees, or rewind and branch a conversation to try another approach. - **Slow work doesn't block you.** Long commands become background jobs, and a finished job wakes the agent, so it can watch tests or CI and fix what fails. - **Agents in parallel.** A git worktree per session lets several agents work on one repo at once. - **Rewind and fork.** `/tree` returns to any point in a conversation and branches from there. Every session is kept, so you can resume or search it. - **Change your view, not your history.** Show or hide commands, output, and diffs throughout the conversation. Hidden output stays available, and `/tools` lets you inspect every tool call and its full output. It runs on any model [Pydantic AI](https://ai.pydantic.dev/) supports, or on your Claude or ChatGPT subscription. **Documentation: [cruxwell.github.io/pcode](https://cruxwell.github.io/pcode/)** **Tried it? [Share feedback](https://github.com/cruxwell/pcode/discussions/4)**: what worked, what didn't, and where you got stuck. ## Install With [Homebrew](https://brew.sh/): ```sh brew tap cruxwell/pcode https://github.com/cruxwell/pcode.git brew install cruxwell/pcode/pcode ``` Or from [PyPI](https://pypi.org/project/pcode/) with [uv](https://docs.astral.sh/uv/): ```sh uv tool install --python 3.15 'pcode[claude]' ``` `brew install --HEAD` tracks `master` instead of the latest release. ## Quick start ```sh pcode -m claude:claude-sonnet-5 # Claude subscription; run /login claude if not signed in pcode -m openai-codex:gpt-5.6-luna # ChatGPT subscription (/login openai-codex) pcode -m anthropic:claude-sonnet-5 # or any provider, with its API key set pcode # reuses the model /model (Ctrl+L) saved; with none, an offline preview pcode -C /path/to/repo "Summarize the open TODOs" git diff | pcode -p --no-save # non-interactive: reply to stdout pcode --continue # resume this directory's newest session pcode --theme-preview # offline sample output and the style gallery ``` **The agent edits files and runs shell commands with your permissions and no approval prompt.** Read [tool permissions](https://cruxwell.github.io/pcode/tools/#tool-permissions) before pointing it at anything you care about. ## More - **Your subscription:** Claude runs through Anthropic's own Agent SDK and Claude Code login (the route Anthropic supports), and ChatGPT through a Codex login. See [providers](https://cruxwell.github.io/pcode/providers/). - [Email remote control](https://cruxwell.github.io/pcode/email/): send a task from Gmail on your phone, reply to keep going, and take the session over at a terminal with `pcode --attach`. - [Custom keybindings and Vim editing](https://cruxwell.github.io/pcode/keybindings/): map keys to commands with arguments, add a Space leader in normal mode, or use `jj` to leave insert mode. - `/btw` side questions, background sessions, recall of past sessions, a browser the agent can drive, and Python extensions. - Built on Pydantic AI's Harness coder, with MCP (OAuth and tool search), searchable sessions, jobs and worktrees on top. Extensions are plain Pydantic AI capabilities. The conversation tree follows [pi](https://github.com/badlogic/pi-mono/blob/main/packages/coding-agent/docs/tree.md)'s. See [PLAN.md](https://github.com/cruxwell/pcode/blob/master/PLAN.md) for the longer-term direction. ## Documentation | Page | What it covers | | --- | --- | | [Getting started](https://cruxwell.github.io/pcode/getting-started/) | Homebrew, uv and source installs, signing in, `--print`, shell completion | | [Scrollback and transparency](https://cruxwell.github.io/pcode/guide/scrollback/) | Guide: what goes into scrollback, `/tools`, `/diffs` | | [A shell for long-running work](https://cruxwell.github.io/pcode/guide/shell/) | Guide: background jobs, watching CI | | [Parallel agents](https://cruxwell.github.io/pcode/guide/parallel/) | Guide: worktrees, parallel sub-agents, `/agents` | | [Extending pcode](https://cruxwell.github.io/pcode/guide/extending/) | Guide: extensions, skills, settings | | [Providers and models](https://cruxwell.github.io/pcode/providers/) | Authentication, supported providers, the model picker, reasoning effort, Claude Code, proxies | | [Configuration](https://cruxwell.github.io/pcode/configuration/) | `pcode config`, per-repository overrides, trusting repository code, the settings table, syntax styles | | [Commands and keys](https://cruxwell.github.io/pcode/commands/) | Slash commands, key bindings, vi mode, tmux newlines, status line, `!command`, the diff and tool inspectors | | [Keybindings](https://cruxwell.github.io/pcode/keybindings/) | Custom command mappings with `/bind`, Ctrl shortcuts, vi editing, a normal-mode leader, and custom escape sequences such as `jj` | | [Terminal setup](https://cruxwell.github.io/pcode/terminal/) | Tab titles, the progress bar, symbols and colors, iTerm2 and tmux settings | | [Tools](https://cruxwell.github.io/pcode/tools/) | Tool permissions, web search, the browser, code mode | | [MCP servers](https://cruxwell.github.io/pcode/mcp/) | Opt-in MCP configuration, OAuth sign-in, deferred tool search | | [Working in a repository](https://cruxwell.github.io/pcode/workspace/) | `AGENTS.md`/`CLAUDE.md`, skills as slash commands, one worktree per session | | [Sessions and recovery](https://cruxwell.github.io/pcode/sessions/) | Saving, resuming, recalling earlier sessions, checkpoints, retries | | [Context, limits and caching](https://cruxwell.github.io/pcode/context/) | Prompt overhead, compaction, output limits, prompt cache notices | | [The transcript](https://cruxwell.github.io/pcode/transcript/) | What lands in scrollback: diffs, thinking, errors, command output, `/redraw` | | [Email remote control](https://cruxwell.github.io/pcode/email/) | Driving a session from Gmail, the remote profile | | [Conversation tree](https://cruxwell.github.io/pcode/conversation-tree/) | `/tree`: rewinding and forking a conversation | | [Side questions](https://cruxwell.github.io/pcode/side-questions/) | `/btw`: asking about the running turn without interrupting it | The same pages build into a browsable site with `make docs-serve`; `make docs` checks every page and anchor link. ## Contributing Bugs: [open an issue](https://github.com/cruxwell/pcode/issues/new/choose). Feedback and ideas: [the feedback discussion](https://github.com/cruxwell/pcode/discussions/4). Security problems: see [SECURITY.md](https://github.com/cruxwell/pcode/blob/master/SECURITY.md). Setup and checks are in [CONTRIBUTING.md](https://github.com/cruxwell/pcode/blob/master/CONTRIBUTING.md). ```sh make test # fast suite; real-tmux regressions skipped make test-all # everything, before touching layout, streaming, the editor, or the prompt ``` [AGENTS.md](https://github.com/cruxwell/pcode/blob/master/AGENTS.md) has the worktree workflow and the traps worth knowing before editing. Contributor notes (architecture, dependencies, profiling, prompt caching, provider design) live in [`dev/`](https://github.com/cruxwell/pcode/tree/master/dev), outside the published docs.