Building in the open
The roadmap.
Tau ships in small, documented phases — each one a readable slice of the agent, not a hidden branch. This page tracks what's shipped and what's next; the reasoning behind each phase lives in dev-notes/.
the frontier
What's next
Extensions marketplace
Up nextDiscover, install, update, and share community-built Tau extensions.
shipped post-roadmap
Since then
Python SDK
A typed Python client for embedding Tau's harness and sessions directly in other Python programs, without going through the CLI.
RPC support
Drive Tau as a subprocess over a structured RPC protocol, so other apps and editors can embed the agent without shelling out to the TUI.
Native local model integration
First-class llama.cpp integration for discovering, configuring, and running local models directly from Tau.
Extensions
Local extension loading; extension-contributed tools, commands, prompt snippets/guidelines; event subscriptions.
System prompt display command
/system shows the effective system prompt Tau will send to the provider, without touching the transcript.
Session tree branching
A /tree modal for branching and revisiting session history, with branch summaries and tool-call filtering.
getting it right
Pre-extension polish
Advanced TUI and product polish
Richer transcript rendering, diff viewer, model/session picker UI, command palette, keybinding customization, theme polish.
Compaction and context management
Token/context estimation, manual /compact, automatic compaction threshold, compaction entries in the session tree.
Session visualization and export
Export a session transcript/tree, human-readable HTML view, preserved branch structure.
Skill invocation reliability
Loaded skills callable by default, manual /skill:<name> triggering, deterministic tests for skill delivery.
Thinking mode controls
Explicit thinking-mode model, TUI control with a Shift-Tab shortcut, active mode shown in the sidebar.
Accurate context accounting and sidebar refresh
Recalculate context/token usage after each turn and keep the TUI sidebar in sync.
before extensions
Hardening
Packaging and installation polish
Installable user-facing tool experience, uv tool/pipx docs, first-run setup docs, format docs.
Project context discovery and reload
Auto-discovery of AGENTS.md and instruction files, /context and /reload commands.
Provider configuration and setup
Durable provider config, multiple OpenAI-compatible providers, default provider/model, setup/login flow.
TUI transcript wrapping and resize behavior
Reflowed transcript text on resize, no horizontal scroll for normal messages, sane handling of long output.
TUI slash-command autocomplete
Autocomplete for commands, /skill:, and prompt templates with Tab/arrow-key selection.
Robust skills and prompt discovery
Load skills/prompts from Tau home, .agents, and project resources with deterministic precedence and diagnostics.
Slash command registry
Registry-based command handling shared by CLI/TUI, built-in commands, extension-ready internal API.
Session manager and resume
Create/list/resume sessions, session metadata index, tau sessions, tau resume <id>, TUI resume/new-session flow.
Tau home, paths, and automatic .agents resources
Canonical Tau home/resource paths, user-level and project-level .agents auto-loading, documented path precedence.
sessions & skills
A real coding agent
Textual TUI foundation
Textual app behind an adapter boundary, TUI state adapter, prompt input/transcript/status, basic commands and cancellation.
Print and event rendering modes
Final text mode, JSON event stream mode, transcript mode, renderer boundary in tau_coding.
System prompt assembly
Pi-style prompt builder, tool snippets/guidelines, skill index, project context, date/cwd suffix.
Skills and prompt templates
Markdown skills, markdown prompt templates, /skill:name expansion, resource path primitives.
Coding session wrapper
CodingSession, persistence around the harness, initial command seam, prompt expansion seam.
Session tree and JSONL persistence
Append-only session entries, JSONL storage, session replay, active leaf reconstruction.
the core loop
Foundation
Non-interactive print-mode CLI
Single prompt CLI, provider/model wiring, built-in tool registration.
Built-in coding tools
read, write, edit, bash — with per-tool prompt snippets and guidelines.
AgentHarness
Transcript owner, prompt()/continue_(), cancellation boundary.
Pure agent loop
Provider stream consumption, assistant message assembly, tool execution loop, agent events.
AI provider layer
Provider interface, fake provider for tests, OpenAI-compatible provider, environment-based API config.
Core types and event model
Provider-neutral messages, tool calls/results, agent events.
Project foundation and design docs
Package scaffold, docs and ADRs, test/lint/format/typecheck setup, basic `tau --version`.
Roadmap phases turn into build journals under dev-notes/, then into these docs. Follow along or open an issue with ideas.